AES

JOURNAL OF THE AUDIO ENGINEERING SOCIETY

AUDIO / ACOUSTICS / APPLICATIONS
Volume 52 Number 4 2004 April

In this issue...
Interpreting Multitone Distortion
Spectrum
Compensating for Lossy Voice Coils
Database Searching to Identify
Audio
Pitch Detection with Percussive
Background
Features...
25th Conference
London—Preview
Technology Overview of
Loudspeakers
DSP in Loudspeakers
Surround Live Summary
26th Conference, Baarn
® Call for Papers

AUDIO ENGINEERING SOCIETY, INC.
INTERNATIONAL HEADQUARTERS
60 East 42nd Street, Room 2520, New York, NY 10165-2520, USA
Tel: +1 212 661 8528
Fax: +1 212 682 0477
E-mail: HQ@aes.org
Internet: http://www.aes.org
ADMINISTRATION
Roger K. Furness Executive Director
Sandra J. Requa Executive Assistant to the Executive Director
OFFICERS 2003/2004
Ronald Streicher President
Theresa Leonard President-Elect
Kees A. Immink Past President
Jim Anderson Vice President
Eastern Region, USA/Canada
Frank Wells Vice President
Central Region, USA/Canada
Bob Moses Vice President
Western Region, USA/Canada
Søren Bech Vice President
Northern Region, Europe
Bozena Kostek
Vice President, Central Region, Europe
Ivan Stamac
Vice President, Southern Region, Europe
Mercedes Onorato Vice President
Latin American Region
Neville Thiele
Vice President, International Region
Han Tendeloo Secretary
Marshall Buck Treasurer
GOVERNORS
Jerry Bruck
Curtis Hoyt
Garry Margolis
Roy Pritts
Don Puluse
Richard Small
Peter Swarte
Kunimaro Tanaka
COMMITTEES
AWARDS
Garry Margolis Chair
CONFERENCE POLICY
Søren Bech Chair
CONVENTION POLICY & FINANCE
Marshall Buck Chair
EDUCATION
Theresa Leonard Chair
FUTURE DIRECTIONS
Ron Streicher Chair
HISTORICAL
J. G. (Jay) McKnight Chair
Irving Joel Vice Chair
Donald J. Plunkett Chair Emeritus
LAWS & RESOLUTIONS
Theresa Leonard Chair
MEMBERSHIP/ADMISSIONS
Francis Rumsey Chair
NOMINATIONS
Kees A. Immink Chair
PUBLICATIONS POLICY
Richard H. Small Chair
REGIONS AND SECTIONS
Subir Pramanik and
Roy Pritts Cochairs
STANDARDS
Richard Chalmers Chair
Correspondence to AES officers and committee chairs should be addressed to them at the society’s international headquarters.
.
.
TELLERS
Christopher V. Freitag Chair
TECHNICAL COUNCIL
Wieslaw V. Woszczyk Chair
Jürgen Herre and
Robert Schulein Vice Chairs
TECHNICAL COMMITTEES
ACOUSTICS & SOUND
REINFORCEMENT
Mendel Kleiner Chair
Kurt Graffy Vice Chair
ARCHIVING, RESTORATION AND
DIGITAL LIBRARIES
David Ackerman Chair
AUDIO FOR GAMES
Martin Wilde Chair
AUDIO FOR
TELECOMMUNICATIONS
Bob Zurek Chair
Andrew Bright Vice Chair
CODING OF AUDIO SIGNALS
James Johnston and
Jürgen Herre Cochairs
AUTOMOTIVE AUDIO
Richard S. Stroud Chair
Tim Nind Vice Chair
HIGH-RESOLUTION AUDIO
Malcolm Hawksford Chair
Vicki R. Melchior and
Takeo Yamamoto Vice Chairs
LOUDSPEAKERS & HEADPHONES
David Clark Chair
Juha Backman Vice Chair
MICROPHONES & APPLICATIONS
David Josephson Chair
Wolfgang Niehoff Vice Chair
MULTICHANNEL & BINAURAL
AUDIO TECHNOLOGIES
Francis Rumsey Chair
Gunther Theile Vice Chair
NETWORK AUDIO SYSTEMS
Jeremy Cooperstock Chair
Robert Rowe and Thomas
Sporer Vice Chairs
AUDIO RECORDING & STORAGE
SYSTEMS
Derk Reefman Chair
Kunimaro Tanaka Vice Chair
PERCEPTION & SUBJECTIVE
EVALUATION OF AUDIO SIGNALS
Durand Begault Chair
Søren Bech and Eiichi Miyasaka
Vice Chairs
SEMANTIC AUDIO ANALYSIS
Mark Sandler Chair
SIGNAL PROCESSING
Ronald Aarts Chair
James Johnston and Christoph M.
Musialik Vice Chairs
STUDIO PRACTICES & PRODUCTION
George Massenburg Chair
Alan Parsons, David Smith and
Mick Sawaguchi Vice Chairs
TRANSMISSION & BROADCASTING
Stephen Lyman Chair
Neville Thiele Vice Chair
STANDARDS COMMITTEE
Richard Chalmers Mark Yonge
Chair Secretary, Standards Manager
John Woodgate
Vice Chair
Yoshizo Sohma
Vice Chair, International
Bruce Olson
Vice Chair, Western Hemisphere
SC-02 SUBCOMMITTEE ON DIGITAL AUDIO
Robin Caine Chair Robert A. Finger Vice Chair
Working Groups
SC-02-01 Digital Audio Measurement Techniques
Richard C. Cabot, I. Dennis, M. Keyhl
SC-02-02 Digital Input-Output Interfacing:
John Grant, Robert A. Finger
SC-02- 05 Synchronization: Robin Caine
SC-03 SUBCOMMITTEE ON THE PRESERVATION AND RESTORATION
OF AUDIO RECORDING
Ted Sheldon Chair Dietrich Schüller Vice Chair
Working Groups
SC-03-01 Analog Recording: J. G. McKnight
SC-03-02 Transfer Technologies: Lars Gaustad, Greg Faris
SC-03-04 Storage and Handling of Media: Ted Sheldon, Gerd Cyrener
SC-03-06 Digital Library and Archives Systems: David Ackerman,
Ted Sheldon
SC-03-12 Forensic Audio: Tom Owen, M. McDermott
Eddy Bogh Brixen
SC-04 SUBCOMMITTEE ON ACOUSTICS
Mendel Kleiner Chair David Josephson Vice Chair
Working Groups
SC-04-01 Acoustics and Sound Source Modeling
Richard H. Campbell, Wolfgang Ahnert
SC-04-03 Loudspeaker Modeling and Measurement
David Prince, Neil Harris, Steve Hutt
SC-04-04 Microphone Measurement and Characterization
David Josephson, Jackie Green
SC-04-07 Listening Tests: David Clark, T. Nousaine
SC-05 SUBCOMMITTEE ON INTERCONNECTIONS
Ray Rayburn Chair John Woodgate Vice Chair
Working Groups
SC-05-02 Audio Connectors
Ray Rayburn, Werner Bachmann
SC-05-05 Grounding and EMC Practices
Bruce Olson, Jim Brown
SC-06 SUBCOMMITTEE ON NETWORK AND FILE TRANSFER OF AUDIO
Robin Caine Chair Steve Harris Vice Chair
Working Groups
SC-06-01 Audio-File Transfer and Exchange
Mark Yonge, Brooks Harris
SC-06-02 Audio Applications Using the High Performance Serial
Bus (IEEE: 1394): John Strawn, Bob Moses
SC-06-04 Internet Audio Delivery System
Karlheinz Brandenburg
SC-06-06 Audio Metadata
C. Chambers
Europe Conventions
AES REGIONAL OFFICES

Zevenbunderslaan 142/9, BE-1190 Brussels, Belgium, Tel: +32 2 345

AES Journal of the Audio Engineering Society
(ISSN 0004-7554), Volume 52, Number 4, 2004 April
7971, Fax: +32 2 345 3419, E-mail for convention information: Published monthly, except January/February and July/August when published bi-
euroconventions@aes.org. monthly, by the Audio Engineering Society, 60 East 42nd Street, New York, New
Europe Services York 10165-2520, USA, Telephone: +1 212 661 8528. Fax: +1 212 682 0477.
B.P. 50, FR-94364 Bry Sur Marne Cedex, France, Tel: +33 1 4881 4632, E-mail: HQ@aes.org. Periodical postage paid at New York, New York, and at an ad-
Fax: +33 1 4706 0648, E-mail for membership and publication sales: ditional mailing office. Postmaster: Send address corrections to AES Journal of the
euroservices@aes.org. Audio Engineering Society, 60 East 42nd Street, New York, New York 10165-2520.
United Kingdom
British Section, Audio Engineering Society Ltd., P. O. Box 645, Slough, The Audio Engineering Society is not responsible for statements made by its
SL1 8BJ UK, Tel: +44 1628 663725, Fax: +44 1628 667002, contributors.
E-mail: UK@aes.org.
Japan EDITORIAL STAFF
AES Japan Section, 1-38-2 Yoyogi, Room 703, Shibuyaku-ku, Daniel R. von Recklinghausen Editor
Tokyo 151-0053, Japan, Tel: +81 3 5358 7320, Fax: +81 3 5358 7328,
E-mail: aes_japan@aes.org. William T. McQuaide Managing Editor Ingeborg M. Stochmal
Gerri M. Calamusa Senior Editor Copy Editor
AES REGIONS AND SECTIONS
Abbie J. Cohen Senior Editor Barry A. Blesser
Eastern Region, USA/Canada Mary Ellen Ilich Associate Editor Consulting Technical Editor
Sections: Atlanta, Boston, District of Columbia, New York, Philadelphia, Toronto Patricia L. Sarch Art Director
Student Sections: American University, Appalachian State University, Berklee Stephanie Paynes
College of Music, Carnegie Mellon University, Duquesne University, Fredonia, Flávia Elzinga Advertising Writer
Full Sail Real World Education, Hampton University, Institute of Audio Research, REVIEW BOARD
McGill University, New York University, Peabody Institute of Johns Hopkins
University, Pennsylvania State University, University of Hartford, University of Ronald M. Aarts Malcolm O. J. Hawksford D. Preis
Massachusetts-Lowell, University of Miami, University of North Carolina at James A. S. Angus Jürgen Herre Derk Reefman
Asheville, William Patterson University, Worcester Polytechnic Institute George L. Augspurger Tomlinson Holman Francis Rumsey
Central Region, USA/Canada Jerry Bauck Andrew Horner Kees A. Schouhamer
Sections: Central Indiana, Chicago, Cincinnati, Detroit, Kansas City,
James W. Beauchamp Jyri Huopaniemi Immink
Nashville, Nebraska, New Orleans, St. Louis, Upper Midwest, West Michigan James D. Johnston Manfred R. Schroeder
Student Sections: Ball State University, Belmont University, Columbia Col- Søren Bech
Arie J. M. Kaizer Robert B. Schulein
lege, Michigan Technological University, Middle Tennessee State University, Durand Begault James M. Kates Richard H. Small
Music Tech College, SAE Nashville, Ohio University, Ridgewater College, Barry A. Blesser
Hutchinson Campus, Texas State University–San Marcos, University of D. B. Keele, Jr. Julius O. Smith III
Arkansas-Pine Bluff, University of Cincinnati, University of Illinois-Urbana- John S. Bradley Mendel Kleiner Gilbert Soulodre
Champaign, University of Michigan, Webster University Robert Bristow-Johnson David L. Klepper Herman J. M. Steeneken
Western Region, USA/Canada John J. Bubbers Wolfgang Klippel John S. Stewart
Sections: Alberta, Colorado, Los Angeles, Pacific Northwest, Portland, Marshall Buck W. Marshall Leach, Jr. John Strawn
San Diego, San Francisco, Utah, Vancouver Stanley P. Lipshitz G. R. (Bob) Thurmond
Student Sections: American River College, Brigham Young University, Mahlon D. Burkhard
California State University–Chico, Citrus College, Cogswell Polytechnical Richard C. Cabot Robert C. Maher Jiri Tichy
College, Conservatory of Recording Arts and Sciences, Expression Center Robert R. Cordell Dan Mapes-Riordan Floyd E. Toole
for New Media, Long Beach City College, San Diego State University, San
Andrew Duncan J. G. (Jay) McKnight Emil L. Torick
Francisco State University, Cal Poly San Luis Obispo, Stanford University, The Guy W. McNally John Vanderkooy
Art Institute of Seattle, University of Colorado at Denver, University of Southern John M. Eargle
D. J. Meares Alexander Voishvillo
California, Vancouver Louis D. Fielder Robert A. Moog Daniel R. von
Northern Region, Europe Edward J. Foster Brian C. J. Moore Recklinghausen
Sections: Belgian, British, Danish, Finnish, Moscow, Netherlands, Mark R. Gander
Norwegian, St. Petersburg, Swedish James A. Moorer Rhonda Wilson
Student Sections: All-Russian State Institute of Cinematography, Danish, Earl R. Geddes Dick Pierce John M. Woodgate
Netherlands, Russian Academy of Music, St. Petersburg, University of David Griesinger Martin Polon Wieslaw V. Woszczyk
Lulea-Pitea
Central Region, Europe COPYRIGHT ONLINE JOURNAL
Sections: Austrian, Belarus, Czech, Central German, North German, Copyright © 2004 by the Audio Engi- AES members can view the Journal
South German, Hungarian, Lithuanian, Polish, Slovakian Republic, Swiss, neering Society, Inc. It is permitted to online at www.aes.org/journal/online.
Ukrainian quote from this Journal with custom-
Student Sections: Aachen, Berlin, Czech Republic, Darmstadt, Detmold, SUBSCRIPTIONS
ary credit to the source.
Düsseldorf, Graz, Ilmenau, Technical University of Gdansk (Poland), Vienna, The Journal is available by subscrip-
Wroclaw University of Technology tion. Annual rates are $190 surface
COPIES
Southern Region, Europe mail, $240 air mail. For information,
Individual readers are permitted to
Sections: Bosnia-Herzegovina, Bulgarian, Croatian, French, Greek, Israel, Ital- contact AES Headquarters.
ian, Portugal, Romanian, Slovenian, Spanish, Serbia and Montenegro, Turkish photocopy isolated ar ticles for
Student Sections: Croatian, Conservatoire de Paris, Italian, Louis-Lumière research or other noncommercial use. BACK ISSUES
Latin American Region Permission to photocopy for internal or Selected back issues are available:
Sections: Argentina, Brazil, Chile, Colombia, Ecuador, Mexico, Peru, personal use of specific clients is From Vol. 1 (1953) through Vol. 12
Uruguay, Venezuela granted by the Audio Engineering (1964), $10 per issue (members),
Student Sections: Del Bosque University, I.A.V.Q., Javeriana University, Los Society to libraries and other users $15 (nonmembers); Vol. 13 (1965) to
Andes University, Orson Welles Institute, San Buenaventura University, Taller registered with the Copyright Clear-
de Arte Sonoro (Caracas)
present, $6 per issue (members), $11
ance Center (CCC), provided that the (nonmembers). For information, con-
International Region base fee of $1 per copy plus $.50 per
Sections: Adelaide, Brisbane, Hong Kong, India, Japan, Korea, Malaysia, tact AES Headquarters office.
page is paid directly to CCC, 222
Melbourne, Philippines, Singapore, Sydney MICROFILM
Rosewood Dr., Danvers, MA 01923,
PURPOSE: The Audio Engineering Society is organized for the purpose USA. 0004-7554/95. Photocopies of Copies of Vol. 19, No. 1 (1971 Jan-
of: uniting persons performing professional services in the audio engi- individual articles may be ordered uary) to the present edition are avail-
neering field and its allied arts; collecting, collating, and disseminating from the AES Headquarters office at able on microfilm from University
scientific knowledge in the field of audio engineering and its allied arts; Microfilms International, 300 North
advancing such science in both theoretical and practical applications; $5 per article.
and preparing, publishing, and distributing literature and periodicals rela- Zeeb Rd., Ann Arbor, MI 48106, USA.
tive to the foregoing purposes and policies. REPRINTS AND REPUBLICATION
ADVERTISING
MEMBERSHIP: Individuals who are interested in audio engineering may Multiple reproduction or republica-
Call the AES Editorial office or send e-
become members of the AES. Information on joining the AES can be found tion of any material in this Journal
at www.aes.org. Grades and annual dues are: Full members and associate mail to: JournalAdvertising@aes.org.
requires the permission of the Audio
members, $95 for both the printed and online Journal; $60 for online Jour- Engineering Society. Permission MANUSCRIPTS
nal only. Student members: $55 for printed and online Journal; $20 for
online Journal only. A subscription to the Journal is included with all member- may also be required from the For information on the presentation
ships. Sustaining memberships are available to persons, corporations, or author(s). Send inquiries to AES Edi- and processing of manuscripts, see
organizations who wish to support the Society. torial office. Information for Authors.
AES
JOURNAL OF THE
AUDIO ENGINEERING SOCIETY
AUDIO/ACOUSTICS/APPLICATIONS
VOLUME 52 NUMBER 4 2004 APRIL
CONTENT
President’s Message .........................................................................................................Ron Streicher 331

PAPERS
Graphing, Interpretation, and Comparison of Results of Loudspeaker Nonlinear Distortion
Measurements........Alexander Voishvillo, Alexander Terekhov, Eugene Czerwinski, and Sergei Alexandrov 332
For loudspeaker nonlinearity, measurement techniques range from single-tone harmonic distortion, which
is easy to interpret but not indicative of performance with music, to reactions to multitone stimuli, which
are hard to interpret but highly informative. Because multitone techniques have the potential to predict the
perception of nonlinearities, the authors focus on various presentation formats and analysis techniques to
make the relevant information in the thousands of intermodulation products accessible and meaningful.

ENGINEERING REPORTS
Impedance Compensation Networks for the Lossy Voice-Coil Inductance of Loudspeaker
Drivers....................................................................................................................W. Marshall Leach, Jr. 358
The high-frequency rise in the voice-coil impedance of a loudspeaker driver produced by lossy voice-coil
inductance can be approximately cancelled by a Zobel network connected in parallel. Such networks
improve performance by presenting purely resistive impedance to the crossover network. Although higher
order networks can be used, a pair of resistors and capacitors is sufficient for typical drivers.
Scalable, Content-Based Audio Identification by Multiple Independent Psychoacoustic
Matching .............................................................................Geoff R. Schmidt and Matthew K. Belmonte 366
A software audio search system, as an analog to text searching, allows a target music sample to be
identified by matching it to a database containing an inventory of reference samples. Rather than rely on
autonomous metadata, the algorithm uses a sequence of vectors based on perceptual attributes. By
iteratively testing a progression of such vectors, the algorithm has the ability to trade accuracy versus
compute time. With increasing storage capacity to hold virtually unlimited quantities of audio data, an
efficient search algorithm is a necessity.
On the Detection of Melodic Pitch in a Percussive Background .........Preeti Rao and Saurabh Shandilya 378
Although many pitch detection algorithms have been proposed over the years, the problem is particularly
difficult when melodic instruments are accompanied by percussive background. The authors propose a
temporal autocorrelation pitch detector motivated by an auditory model that attempts to suppress errors
produced by inharmonic interfering partials of such instruments as a kick drum. Separate processing of
frequency channels proved crucial in reducing the distortion products, due to the nonlinear hair-cell model,
between the signal harmonics and the interfering partials.

STANDARDS AND INFORMATION DOCUMENTS

AES Standards Committee News........................................................................................................... 392
New AESSC chairman; audio file interchange; audio over IEEE 1394

FEATURES
25th Conference Preview, London ......................................................................................................... 402
Calendar.............................................................................................................................................. 404
Program .............................................................................................................................................. 405
Registration Form .............................................................................................................................. 411
Historical Perspectives and Technology Overview of Loudspeakers for Sound Reinforcement
............................................................................................................................J. Eargle and M. Gander 412
DSP in Loudspeakers.............................................................................................................................. 434
Surround Live Summary ...............................................................................................Frederick Ampel 440
26th Conference, Baarn, Call for Papers ............................................................................................... 457

DEPARTMENTS
Review of Acoustical Patents ...........................395
News of the Sections.........................................443
Upcoming Meetings ..........................................447
Sound Track .......................................................448
New Products and Developments ....................449
Available Literature............................................450
Membership Information...................................451
Advertiser Internet Directory............................453
In Memoriam ......................................................456
Sections Contacts Directory ............................458
AES Conventions and Conferences ................464
 PRESIDENT’S MESSAGE
DISTINGUISHED SPEAKERS PROGRAM
It is with great pleasure that I am able to announce two
initiatives recently approved by the Board of Governors.
The first is the Distinguished Speakers Program, which
will establish a roster of touring guest lecturers who will
be available to provide presentations at the meetings of
local sections throughout the Society. To begin the
program I would ask that anyone who travels regularly on
business and has the time, interest, and ability to offer a
technical presentation to one or more local sections
contact the AES Headquarters office at speakers@aes.org
and let us know your schedule and itinerary. We will then
work to connect you to the appropriate section officers.
Of course, the greater advance notice you can provide of
your planned trip the better we will be able to make these
arrangements to suit your schedule.
It is our goal that this project will develop to provide a
wealth of interesting meetings for all of our local
sections, large and small, throughout the Society,
augmenting their activities and stimulating increased
membership in the Society.
To support the Distinguished Speakers Program, the
Executive Committee has established a special fund that
we hope ultimately will be sufficiently endowed to be
fully self-supporting. To initiate this fund, I have made a
significant personal contribution, and I invite you, either
individually or corporately, to join me in supporting this
project. Contributions are tax deductible and will be used
solely to finance the activities of this program.
To find out more about the Distinguished Speakers
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April
Program, to offer your services as a guest lecturer, or to
make a contribution in support of the program,
please contact the AES Headquarters office at
speakers@aes.org. You can also find out more about the
program at www.aes.org/info/speakers.
LEGACY FUND
Another new initiative is being developed to improve the
long-term financial stability of the AES. Called the
Legacy Fund, this provides an opportunity for members
to make tax-deductible contributions at any time or to
leave bequests to the AES in their wills; similar programs
are used by other professional societies. This endowment
fund will support the general operating budget and activi-
ties of the Society and lessen the need to increase dues or
other membership fees.
As your president for this year, it is my hope that both of
these initiatives will help our Society grow and prosper
long into the future. To accomplish this, however, we need
qualified people to join the Distinguished Speakers
Program roster, and we need your financial support and
contributions to help us develop and sustain these and all of
the activities of the AES.
Ron Streicher, President, rds@aes.org
331
 PAPERS

Graphing, Interpretation, and Comparison of

Results of Loudspeaker Nonlinear
Distortion Measurements*

ALEXANDER VOISHVILLO, AES Member, ALEXANDER TEREKHOV, EUGENE CZERWINSKI,

AND SERGEI ALEXANDROV**

Cerwinski Laboratories Inc., Simi Valley, CA 93063, USA

Harmonic distortion and total harmonic distortion may not convey sufficient information
about nonlinearity in loudspeakers and horn drivers to judge their perceptual acceptability.
Multitone stimuli and Gaussian noise produce a more informative nonlinear response. The
reaction to Gaussian noise can be transformed into coherence or incoherence functions. These
functions provide information about nonlinearity in the form of “easy-to-grasp” frequency-
dependent curves. Alternatively, a multitone stimulus generates a variety of “visible” har-
monic and intermodulation spectral components. If the number of input tones is significant,
the nonlinear reaction may consist of hundreds, if not thousands, of distortion spectral
components. The results of such measurements are difficult to interpret, compare, and over-
lay. A new method of depicting the results of multitone measurements has been developed.
The measurement result is a single, continuous, frequency-dependent curve that takes into
account the level of the distortion products and their “density.” The curves can be easily
overlaid and compared. Future developments of this new method may lead to a correlation
between curves of the level of distortion and the audibility of nonlinear distortion. Using
nonlinear dynamic loudspeaker models, multitone and Gaussian noise test signals are com-
pared with traditional and nontraditional measurement techniques. The relationship between
harmonics and intermodulation products in static and dynamic nonlinear systems is analyzed.

0 INTRODUCTION and harmonic products are characterized by their indi-

Loudspeakers and horn drivers are complex nonlinear
dynamic systems whose nonlinear reactions to different
stimuli may be significantly different. It has been demon-
strated earlier that traditional methods of measuring non-
linear distortion using a sweeping tone, and such criteria as
harmonic distortion and total harmonic distortion (THD),
may not convey sufficient information about the nonlinear
properties of loudspeakers and horn drivers [1]. It has been
demonstrated by using the multitone stimuli that the in-
termodulation products outweigh the harmonics in even
comparatively simple nonlinear systems characterized by
static polynomial nonlinearity. This difference between
harmonics and intermodulation products is especially pro-
nounced if a higher order nonlinearity occurs. In complex
nonlinear dynamic systems, such as electrodynamic direct-
radiating loudspeakers or horn drivers, the intermodulation
*Manuscript received 2002 May 9; revised 2003 December 12
and 2004 February 13.
**Now with JBL Professional, Northridge, CA 91329, USA.
332
vidual dependence on frequency. These individual depen-
dencies prevent a substitution of intermodulation testing
with harmonic measurements.
Complex signals such as multitone stimuli and Gaussian
noise produce nonlinear reactions that carry more infor-
mation about intermodulation of various kinds and orders
in loudspeakers and horn drivers. Nonlinear reactions to
the Gaussian noise can be transformed into, for example,
the Wiener kernels [2], coherence or incoherence func-
tions [3], [4], or the higher order spectra (HOS) [5]. Wie-
ner kernels do not have an intuitive simple interpretation.
The coherence (or incoherence) functions and HOS pro-
vide information about the overall nonlinear behavior of a
measured object in the form of “easy-to-grasp” frequency-
dependent curves. However, these functions are also sen-
sitive to noises and other effects (such as reflections) that
may “mar up” the results of nonlinear testing. In addition,
it takes a comparatively long time to measure them. His-
torically the coherence function has been used in the test-
ing of hearing aids [3], [4] and in the evaluation of non-
linearity and noises in magnetic recording [6], but has
J. Audio Eng. Soc., Vol. 52, No. 4, April 2004
PAPERS COMPARISON OF LOUDSPEAKER NONLINEAR DISTORTION MEASUREMENTS
never been popular in the assessment of loudspeaker non-
linearity. The multitone stimulus has been gaining popu-
larity in many applications during the last decade [7]–[12].
Various aspects of using multitone signals in loudspeaker
testing will be discussed in Section 3. The multitone
stimulus produces a rich spectrum of distortion products.
The statistical distribution and crest factor of multitone
signals is close to that of a musical signal. However, the
results of a measurement presented in the form of an out-
put spectrum are difficult to interpret, compare, and overlay.
A new method of depicting the results of multitone
measurements has been developed in this work. According
to this method, the result of the measurement is presented
as a single, continuous, frequency-dependent curve that
takes into account not only the distortion level of the spec-
tral components but also their “density.” Many such
curves, corresponding to different levels of input signal,
can be overlaid easily in a way that is practically impos-
sible using “unprocessed” responses to the multitone
stimulus. These two- or three-dimensional graphs can eas-
ily demonstrate how the overall nonlinear distortion in a
device measured (loudspeaker or horn driver) increases
with the level of the input multitone stimulus.
In the technical publications of previous years other
approaches to measure, model, and assess loudspeaker
nonlinearity have been discussed. Some of these methods
have not been included in the existing standards and prob-
ably never will. However, a comparative survey helps us
to look at the problem of loudspeaker nonlinearity mea-
surements from a systematic standpoint and to understand
better their meaning, advantages, and limitations. The tra-
ditional and nontraditional methods that will be compared
to the methods based on the application of Gaussian noise
(incoherence function) and multitone stimulus are re-
viewed hereafter.
One of the unconventional methods is the measurement
and graphing of high-order frequency response functions
(HFRFs) derived from the second- and third-order Volt-
erra time-domain kernels. The HFRFs are three-
dimensional graphs, representing a “surface” of distortion
products. Volterra series expansion stems from the funda-
mental theoretical input of Volterra [13] and has been
introduced by Wiener for the analysis of weakly nonlinear
systems, characterized by low levels of distortion (see [14]
for a history of this subject). Since then Volterra series
expansion has been used widely in many areas where the
structure of weakly nonlinear systems is not known, or the
parametric analysis of their behavior is too complicated.
Kaizer pioneered the use of Volterra series in loudspeaker
nonlinear analysis. He derived explicit expressions for
HFRFs through loudspeaker excursion-dependent param-
eters [15]. Kaizer’s research was followed by a number of
works (for example, [16]–[18]), where the second- and
third-order kernels were measured, transformed into cor-
responding HFRFs, and then plotted as three-dimensional
graphs depicting loudspeaker second- and third-order dis-
tortions. Advantages and drawbacks of Volterra series ex-
pansion will be discussed in the Section 3.
Two-tone intermodulation distortion of the second and
third order has traditionally been measured to assess non-
J. Audio Eng. Soc., Vol. 52, No. 4, April 2004
linearity in audio equipment since two methods were in-
troduced in the 1940s by Hilliard [19] (SMPTE inter-
modulation distortion) and Scott [20] (difference-
frequency or CCIF distortion). The former uses one
sweeping and one stationary low-frequency tone of four
times higher amplitude. The latter method uses two closely
spaced simultaneously swept tones. Products of the kind
Pf2±f1 and Pf2±2f1 are plotted. The frequency f1 corresponds
to the fixed tone in the SMPTE method or to the lower
frequency tone in the CCIF method, and f2 is the frequency
of the higher sweeping tone. Both methods do not measure
intermodulation products of order higher than three.
For the measurement of loudspeaker nonlinearity AES
standard AES2-1984 [21] recommends a measurement of
only the second- and third-order harmonic distortion. IEC
standard 60268-5 [22] recommends that a wider circle of
characteristics be measured, including THD, individual
second and third harmonics, and individual second-order
difference intermodulation products. In addition this stan-
dard recommends the aggregated criteria of sound pres-
sure level (SPL) intermodulation in the form (Pf2+f1 +
Pf2−f1)/Pf2 and (Pf2+2f1 + Pf2−2f1)/Pf2, f2 >> f1, where the sum
and difference products of similar order (second or third
only) are summed and related to one of two primary tones.
Alternative approaches to measure loudspeaker inter-
modulation distortion were proposed by Keele [23]. Keele
recommends two methods for consideration, one based on
the use of two-tonal signals, 40 and 400 Hz, of equal
amplitude. The percent of distortion is to be plotted as a
function of input power. The other method includes a
fixed-frequency upper range signal coupled with a swept
bass signal. Keele also advocates the use of the shaped
tone burst for the assessment of loudspeaker maximum
SPL [24].
These various methods and signals provide different
information about the nonlinearity in a measured loud-
speaker. Nevertheless, the following questions remain
open: “What method conveys most adequate information
about nonlinearity in a measured loudspeaker?”, “How
well are the measurement data related to the perceived
deterioration of sound quality or to the malfunctioning of
a loudspeaker?”, and “How can these data be represented
in the most comprehensible manner?”.
This work is intended to illustrate and compare several
methods of assessment and graphical presentation of weak
nonlinearity in loudspeakers. The comparison is carried
out using a nonlinear dynamic model of a low-frequency
loudspeaker that includes excursion-dependent param-
eters: Bl product, suspension stiffness, voice-coil induc-
tance, parainductance, eddy currents–caused resistance,
and voice-coil current-dependent magnetic flux modula-
tion. The models of three different woofers are used for
comparison: 8-in (203-mm) diaphragm long voice coil,
8-in (203-mm) diaphragm short voice coil, and 12-in (305-
mm) diaphragm long voice coil. The measurement results
are simulated at different signal levels. A comparison is
made of THD, harmonic distortion, Volterra second-order
frequency-domain kernels, also called high-order fre-
quency response functions (HOFRF), two-tone sum and
difference intermodulation distortion, two-tone total non-
333
VOISHVILLO ET AL.
linear distortion, multitone intermodulation and multitone
total nonlinear distortion (MTND), and incoherence
function.
1 TEST SUBJECT, TEST RESULT
PRESENTATION, AND GOAL
The possible uncertainty in the objective assessment of
nonlinear systems by traditional testing methods stems
from the complex nature of nonlinear systems. A linear
time-invariant system with a single input and output is
fully described by its pulse response (or by its complex
transfer function). The output signal can be calculated by
the convolution of the input signal with the impulse re-
sponse (in the time domain), or by multiplication of the
input spectra by the complex transfer function (in the fre-
quency domain). In addition, the relationship between an
input and an output signal of a linear system can also be
expressed in the form of a linear differential equation or a
linear difference equation if a linear system is discrete.
Simply speaking, a linear system does not add new fre-
quency components to the output signal.
The behavior of a nonlinear system is substantially
more complex. Traditional methods used for the analysis
of linear systems are not applicable for an analysis of even
weakly nonlinear systems. The properties of such systems
can be described in the time domain by the sum of Volt-
erra kernels [14]. The latter are essentially the pulse re-
sponses responsible for the transformation of input signal
by nonlinearities of different orders. The overall pulse
response of a weakly nonlinear system is the sum of ker-
nels of different orders that are multidimensional functions
of time. For example, the pulse response of a simple non-
linear system characterized by a second-order dynamic
weak nonlinearity is the sum of the kernels of the first
order (which is essentially the linear pulse response) and
a second-order kernel. The latter can be presented graphi-
cally as a three-dimensional surface with two horizontal
time scales.
The output of such a system can be expressed as the
convolution of an input signal with the first- and second-
order kernels. This convolution is expressed in general by
multiple integrals. The multidimensionality is also valid
for a frequency-domain complex transfer function of non-
linear systems. The amplitude and phase frequency re-
sponses of a second-order distortion are also three-
dimensional surfaces having two horizontal frequency
scales. The second harmonic distortion response (ampli-
tude and phase) is merely a diagonal “cut” through these
two surfaces. Similarly, the impulse response of the sec-
ond harmonic is merely a diagonal cut across the surface
of the three-dimensional kernel of the second order [1]. It
is obvious that neither the frequency response of the sec-
ond harmonic nor its impulse response will represent the
entire second-order nonlinear response of a weakly non-
linear dynamic system legitimately. These cuts may not
correspond to the maxima of the distortion surface. Using
only harmonic distortion may cause mistakes in the as-
sessment of the nonlinearity. Therefore a search for a cor-
relation between the audibility of nonlinearly distorted
334
PAPERS
musical signals and the level of harmonic distortion can
lead to wrong conclusions. The present example considers
the three-dimensional representation of the second-order
nonlinearity. The responses of the higher order nonlineari-
ties are multidimensional functions. A real dynamic non-
linearity existing in loudspeakers and horn drivers is sig-
nificantly more complex than this simple example.
Imagine a hypothetical loudspeaker whose amplitude
frequency response is presented by only a few samples at
a few frequencies. If there is no information about the
behavior of the amplitude frequency response between
these sparse samples, we cannot make a judgment about its
performance. The response of this loudspeaker might be
perfectly flat between available samples; it might as well
have a strong irregularity. Similarly, a single frequency
response of nonlinear distortion, be it a harmonic or an
intermodulation curve, conveys only limited information
about the nonlinearity. If there is no information about a
surface of nonlinear responses between the available cuts
of harmonic or intermodulation frequency responses, the
behavior of the nonlinear system cannot be assessed ac-
curately. This statement is valid for loudspeakers and horn
drivers, which are complex, dynamic nonlinear systems
with many degrees of freedom and whose nonlinear re-
sponses depend strongly, and in a complex manner, on the
frequency. In amplifiers, for example, the nonlinear char-
acteristics do not exhibit that strong a frequency depen-
dence. Therefore, in their analysis the relationship be-
tween harmonic and intermodulation distortions might be
more predictable.
The examples with nonlinear distortion in the loud-
speakers described in this work will assume weak non-
linearity (distortion products are at least 20–30 dB lower
than the fundamental signal). In reality, however, the dis-
tortion in loudspeakers and horn drivers can be higher,
placing loudspeakers and drivers in the category of
strongly nonlinear systems. These systems are character-
ized by even more sophisticated properties that may in-
clude bifurcation and chaotic and stochastic behavior. This
class of nonlinear systems will not be considered in the
current work.
There is a dilemma in measuring, graphing, and inter-
preting nonlinear distortion. On the one hand the assess-
ment of nonlinear distortion needs the analysis of much
more information than is required to assess a linear sys-
tem. On the other hand this information should be pre-
sented in a simple and comprehensible graphical manner.
These two requirements may contradict each other. Fur-
thermore, the graphed data should be pertinent from the
standpoint of distortion audibility. The final goal of a loud-
speaker nonlinear distortion measurement is to obtain data
that convey adequate information about the nonlinearity so
that this information can be related unambiguously to the
perceived sound quality of a loudspeaker under test, and
that thus the performance of different loudspeakers can be
compared objectively. The measurement data must be
“manageable.” In spite of the seeming simplicity of these
goals, and a nearly 90-year history of numerous efforts of
many researchers (see [1] for a history of the subject),
these goals have never been fully achieved.
J. Audio Eng. Soc., Vol. 52, No. 4, April 2004
PAPERS COMPARISON OF LOUDSPEAKER NONLINEAR DISTORTION MEASUREMENTS

2 PSYCHOACOUSTICAL CONSIDERATIONS stronger signal, called masker. The masking may be ob-
served in the time domain in the form of post and pre-
The search for a correlation between an objective mea- masking when a stronger short-term masker “obliterates” a
surement of nonlinear distortion and the subjective audi- weaker masked signal, even if the latter precedes the
bility of nonlinear distortion in audio equipment in gen- masker. Masking may also occur in the frequency domain,
eral, and in loudspeakers in particular, has always been where a stronger masker produces a shadow zone around
and remains the Holy Grail of the audio industry. Loud- itself. This shadow psychoacoustically suppresses those
speaker distortion measurement data related to the per- masked signals whose spectrum components happen to be
ceived sound quality must not only have some readily within the spectrum and below the level of the masking
comprehensible interpretation, but must also be supported frequency-domain curve. The masking frequency-domain
psychoacoustically. There must be a credible knowledge curve produced by a single tone, for example, resembles a
relating the graphically presented objective data to the triangle. With an increase in the level of the masker, the
subjectively perceived sound quality. Due to the complex triangle becomes asymmetrical, with its longer side
nature of the nonlinearity and the intricacy of the human stretching toward high frequencies [25]. With an increase
auditory system’s reaction to a musical signal adversely in the level of the masking tone the level of the masking
affected by the nonlinearity, there are no undisputedly asymmetrical triangle increases and stretches over a wider
credible and commonly recognized thresholds expressed frequency range, producing a stronger masking effect
in terms of the traditional nonlinear distortion measures above the frequency of the sinusoidal masker rather than
related to the perceived sound quality. The problem is below it (Fig. 1). The masker shown in Fig. 1 corresponds
aggravated by the fact that the objective measurement of to curve a.
nonlinearity deals merely with the symptoms of a nonlin- The asymmetrical triangular shape of the masking curve
ear system, that is, with the reactions of a nonlinear sys- explains why the higher order harmonics and intermodu-
tem, such as a loudspeaker, to various testing signals. Here lation products are more audible than the lower order ones,
we operate with objective categories, such as measured who are more prone to be masked. In Fig. 2 the harmonics
levels, responses, characteristics, and parameters. Mean- and intermodulation products, produced by a two-tone sig-
while the subjective assessment of musical signals im- nal affected by the static fifth-order nonlinearity, are over-
paired by the nonlinearity deals with the human psycho- laid with the masking curve produced by the two-tone
acoustical reactions and impressions expressed in a quite masker. This also explains why the difference intermodu-
different vernacular, such as “acceptable, annoying, pleas-
ant, or irritating.” The objective of a researcher is to put a
bridge between these two different domains.
The dynamic reaction of a complex nonlinear system
(such as a direct-radiating loudspeaker or a horn driver) to
a musical signal cannot be extrapolated from its reaction to
a simple testing signal such as a mere sweeping tone.
Hence the credible thresholds of subjectively perceived
nonlinear distortion expressed in terms of the reaction to
simple sinusoidal signals (THD, harmonics, or two-tone
intermodulation distortion) may not be valid. More com-
plex signals, such as a random or pseudorandom noise or
a multitone stimulus, are believed (by the authors) to be
required to search for subjectively relevant thresholds. Fig. 1. Masking curves corresponding to levels a–e of sinusoidal
The complex properties of the human hearing system, tone masker. The masker corresponds to curve a.
which is a far cry from a mere Fourier frequency analyzer,
only add complexity to the problem. The behavior of the
hearing system is characterized by many effects described
in various publications on psychoacoustics (see [25], for
example). The properties of the auditory system most rel-
evant to the subject of this work are the intrinsic nonlin-
earity of the hearing system and temporal and frequency-
domain masking. These effects have been treated in detail
in the psychoacoustical literature, and it is not the authors’
goal to replicate these texts. However, it is worth men-
tioning that the intrinsic nonlinearity of the human hearing
system manifests itself at high levels of sound pressure,
whereas the masking is a general property of the hearing
system, “working” at any level of the sound pressure signal.
The masking plays a crucial role in the perception of Fig. 2. Masking effects produced by two closely spaced funda-
nonlinear distortion. The crux of masking is a psycho- mental tones (maskers) on their harmonics and intermodulation
acoustical suppression of a weaker masked signal by a products.

J. Audio Eng. Soc., Vol. 52, No. 4, April 2004 335

VOISHVILLO ET AL.
lation products are more susceptible to be outside the nar-
rower lower side of the masking curve, which makes them
more audible. In addition, the frequencies of the harmon-
ics have a higher probability of coinciding with overtones
of particular musical instruments, and of being masked by
these overtones as well. Meanwhile the variety of “disso-
nant” intermodulation products of various orders do not
coincide with the overtones of the musical instruments
and, therefore, are more noticeable.
The complexity of nonlinear systems such as loud-
speakers and the complexity of the hearing system explain
why thresholds of distortion audibility expressed in terms
of such plain metrics as harmonic distortion, THD, and
two-tone intermodulation distortion strongly depend on
the type of musical signal used in the experiments, and
why the data obtained by different researchers are often
inconsistent. An historical review of the search for a rela-
tionship between objective data (expressed in terms of
such metrics as THD, harmonic distortion, and two-tone
intermodulation distortion) and the subjectively perceived
deterioration of the sound quality of reproduced material is
given in [1]. Historically many early research works in this
area did not have a clear understanding of the complex
nature of nonlinear dynamic systems and suffered from a
lack of the modern knowledge of the principles of opera-
tion of the hearing system. Since then the theory of non-
linear systems and the knowledge of psychoacoustics have
progressed enormously. Examples of the use of this prog-
ress are the systems of low-bit compression (such as MP3,
WMA, ATRAC) that “deceive” the hearing system by
deleting significant parts of signal information without a
significant deterioration of the perceived sound quality.
These low-bit-rate compression systems, evaluated by
standard metrics such as THD or two-tone intermodulation
distortion, would have exhibited unacceptable levels of
distortion, proving that the standard metrics have no im-
mediate relationship with the perceived sound quality.
Continuing this line of thought, THD, which is the most
popular measure of nonlinear distortion in audio, is not a
reliable measure of the psychoacoustically meaningful
nonlinearity in a loudspeaker. First, it does not add any-
thing to what individual harmonic curves can show. Sec-
ond, since useful information about harmonics of different
orders is not available from THD, its interpretation may
result in wrong conclusions about the character of the
nonlinearity in a loudspeaker tested. In other words, the
same 10% THD of the sound pressure level at a certain
level of input voltage might be produced by the dominant
second- and third-order nonlinearities in one loudspeaker,
or it might include the higher order harmonics in another
loudspeaker as well. The difference in the amount and
level of intermodulation products and, correspondingly, in
the sound quality of these two loudspeakers could be sig-
nificant. This THD would not indicate.
As has been mentioned, the multitone stimulus, whose
objective parameters, such as the probability density func-
tion, have similarity with a musical signal, seems to be a
good candidate for a better testing signal. However, there
is an important aspect of using multitone stimuli that
should be considered here to be objective. Currently the
336
PAPERS
interpretation of multitone test results does not have a well
substantiated psychoacoustical support. So far we cannot
derive precise judgments about the sound quality of a
loudspeaker (that has been tested by a multitone stimulus)
from the response to this signal. However, the results of
recent research on the correlation between objective mea-
surements and subjectively perceived nonlinearly distorted
speech and musical signals [26] prove that for certain
kinds of nonlinearity the postprocessed reaction to a mul-
titone stimulus expressed as a single number, dubbed by
the authors of that work the distortion score (DS), has a
very high correlation with subjectively perceived sound
quality. The distortion score is obtained by the summation
of the levels of distortion products within the mean
equivalent rectangular bandwidth (ERBN) of the auditory
filter, which is conceptually similar to the traditional criti-
cal bandwidth but differs in numerical values. It is be-
lieved that future experiments with multitone stimuli
might lead to further positive results in attempts to find a
relationship between the objective measurement data and
subjectively perceived sound quality.
3 TESTING METHODS AND INTERPRETATION
OF MEASUREMENT DATA
3.1 Relationship between Harmonics and
Intermodulation Products—Effects Produced by
Static Nonlinearity of Different Orders
Measurements of nonlinear distortion using simple ex-
citation signals may not provide adequate information
about the nonlinear properties of a device under test. Even
considering a simple form of static nonlinearity, some not
immediately obvious effects appear. Let a hypothetical
static nonlinear system be governed by the simple poly-
nomial expression
n
y共t兲 = 兺h z 共t兲
i=0
i
i
(1)
where z(t) is an input signal, y(t) is an output signal, h0 is
the dc distortion component, h1 is the linear gain coeffi-
cient, and h2, . . . , hn are the weighting coefficients re-
sponsible for the influence of a nonlinearity of a particular
order beginning from the second. The coefficients hi in
general may have positive or negative signs, and some of
them may be zero.
A nonlinearity of this kind might, for example, approxi-
mate a loudspeaker suspension in the form of a relation-
ship between the diaphragm displacement x and the force
F if creep effect (the long-term dependence of the com-
pliance on the time of loading) and hysteresis are omitted.
Then the coefficients h0, . . ., hn in Eq. (1) represent the
suspension compliance. As a loudspeaker operates, non-
linear compliance causes nonlinear displacement, and this
effect interacts with other nonlinear phenomena. The over-
all nonlinearity of loudspeakers is dynamic and more com-
plex than the simple relationship described by Eq. (1).
Bearing in mind that this particular example is not a com-
plete representation of the operation of a loudspeaker, we
will nevertheless analyze this simple static nonlinearity to
illustrate some general effects.
J. Audio Eng. Soc., Vol. 52, No. 4, April 2004
PAPERS COMPARISON OF LOUDSPEAKER NONLINEAR DISTORTION MEASUREMENTS
Let us assume that the relationship between the dis-
placement, x (output) and the force F (input) is described
by the expression: x(t) ⳱ c1F(t) − c5F5(t), where the co-
efficients c1 and −c5 represent nonlinear mechanical com-
pliance. We also assume that the input driving force is
sinusoidal and the coefficient c5 is set to c5 ⳱ 0.26344 to
produce 10% THD in displacement. Fig. 3 shows the de-
pendence of the displacement on the driving force and the
spectrum of displacement corresponding to the sinusoidal
input. The spectral components of displacement are de-
scribed by the expression
x共t兲 = c1F sin ␻t − c5F 5 sin5 ␻t
= c1F sin ␻t − c5F 5共0.625 sin ␻t − 0.3125 sin 3␻t
+ 0.0625 sin 5␻t兲. (2)
The fifth-order “limiting” nonlinearity produces the
fifth harmonic (which is quite predictable). It also gener-
ates the third harmonic and the spectral component having
the same frequency as the input signal. Since the latter
spectral component is out of phase with the fundamental
tone it produces the limiting effect of the suspension be-
cause it decreases the level of the first harmonic in the
displacement compared to the linear one. The fifth har-
monic is five times (−14 dB) smaller than the third and ten
times (−20 dB) smaller that the spectral component having
the same frequency as the input tone.
Fig. 3. (a) Dependence of suspension displacement on force;
fifth-order approximation. (b) Spectrum of displacement corre-
sponding to fifth-order approximation. Sinusoidal input.
J. Audio Eng. Soc., Vol. 52, No. 4, April 2004
If the input force F has the form of a two-tone signal,
F ⭈ 0.5(sin ␻1t + sin ␻2t), then the output signal (displace-
ment) consists of the linear part c1F ⭈ 0.5(sin ␻1t + sin ␻2t)
and of the distortion products generated by the fifth-order
nonlinearity, which include two fifth and two third har-
monics, and twelve intermodulation products. In addition,
the fifth-order nonlinearity produces two spectral compo-
nents having frequencies identical to the frequencies of the
initial input signals. Fig. 4 depicts the output spectrum. We
assume that the amplitude of each tone is half the ampli-
tude of the previous sinusoidal tone to maintain the same
maximum level as the single-tone signal,
x共t兲 = c1F ⭈ 0.5共sin ␻1t + sin ␻2t兲
− c5F 5 ⭈ 0.55共sin ␻1t + sin ␻2t兲5
= c1F ⭈ 0.5共sin ␻1t + sin ␻2t兲
− c5F 5 ⭈ 0.03125关0.0625 sin 5␻1t
+ 0.0625 sin 5␻2t − 1.5625 sin 3␻1t
− 1.5625 sin 3␻2t + 6.25 sin ␻1t + 6.25 sin ␻2t
− 3.125 sin 共2␻1 + ␻2兲t + 3.125 sin 共2␻1 − ␻2兲t
− 3.125 sin 共2␻2 + ␻1兲t + 3.125 sin 共2␻2 − ␻1兲t
+ 0.3125 sin 共4␻1 + ␻2兲t − 0.3125 sin 共4␻1 − ␻2兲t
+ 0.3125 sin 共4␻2 + ␻1兲t − 0.3125 sin 共4␻2 − ␻1兲t
+ 0.625 sin 共3␻1 + 2␻2兲t + 0.625 sin 共3␻1 − 2␻2兲t
+ 0.625 sin 共3␻2 + 2␻1兲t + 0.625 sin 共3␻2 − 2␻1兲t兴.
(3)
The balance between fifth and third harmonics becomes
significantly different compared to the single-tone excita-
tion. The fifth harmonic turns out to be much lower in
amplitude than the third harmonic and all intermodulation
products. The difference between the fifth and third har-
monics produced by the same fifth-order nonlinearity be-
comes 28 dB. All twelve intermodulation products are
higher in amplitude than the fifth harmonic. If the maxi-
mum level of the two-tone signal is chosen equal to the
amplitude of the single-tone signal producing 10% THD,
the relationship between the harmonics produced by these
two signals is as shown in Table 1.
From this observation it might follow that if someone
tests only the harmonic distortion in this hypothetical non-
linear suspension, he might come to a conclusion that this
suspension is impaired predominantly by the third har-
monic distortion and to a lesser degree by the fifth har-
Fig. 4. Spectrum of nonlinear reaction to two-tone input; fifth-
order approximation.
337
VOISHVILLO ET AL.
monic distortion. This conclusion might lead to a “picture”
of a musical signal contaminated predominantly with third
and slightly with fifth-order harmonic distortion. How-
ever, the application of a more complex two-tone signal
changes the reaction of this nonlinear system. For ex-
ample, the fifth harmonic becomes very small and virtu-
ally unessential compared to other distortion components.
In its turn, the power of the stronger third harmonic be-
comes negligible compared to the power of the intermodu-
lation products. This observation can be extrapolated eas-
ily to higher order static nonlinearities.
This simple example illustrates the fact that harmonics
and intermodulation products are merely symptoms of
nonlinearity. They are the reaction of a nonlinear system to
a particular input signal. Different inputs produce different
symptoms in the same nonlinear system. Since a real mu-
sical signal is a set of various spectral components rather
than a merely sinusoidal tone, we can assume that the
share of harmonics is much less significant than the share
of intermodulation products if a musical signal is applied
to a static nonlinear system. The relationship between har-
monics and intermodulation products in dynamic nonlin-
ear systems will be analyzed in detail in the next sections.
The preceding example illustrated the different reac-
tions of the same static nonlinear system to different sig-
nals. Now let us consider the role played by the static
nonlinearity of different orders. Let the hypothetical non-
linear suspension be approximated by the expression x(t)
⳱ c1F(t) − c3F3(t) and let the sinusoidal signal be applied
to it,
x共t兲 = c1F sin ␻t − c3F 3 sin3 ␻t
= c1F sin ␻t − c3F 3共0.75 sin ␻t − 0.25 sin 3␻t兲.
Single Tone
Harmonics
1st 3rd 5th
−1.6 dB −21.7 dB −35.7 dB
Fig. 5. (a) Dependence of suspension displacement on force; third-order approximation. (b) Spectrum of displacement corresponding
to third-order approximation. Sinusoidal input.
338
PAPERS
Fig. 5 illustrates the dependence of displacement on the
driving force and the spectrum of displacement corre-
sponding to sinusoidal input.
A sinusoidal input obviously produces third harmonics.
It also produces the spectral component of the “first or-
der,” which has the same frequency as the input signal. If
a two-tone signal is applied to the same system, it pro-
duces four intermodulation products, two third harmonics,
and two terms of the “first order,”
x共t兲 = c1F ⭈ 0.5共sin ␻1t + sin ␻2t兲
− c3F 3 ⭈ 0.53共sin ␻1t + sin ␻2t兲3
= c1F ⭈ 0.5共sin ␻1t + sin ␻2t兲
− c3F 3 ⭈ 0.125关−0.25 sin 3␻1t − 0.25 sin 3␻2t
+ 2.25 sin ␻1t + 2.25 sin ␻2t
− 0.75 sin 共2␻1 + ␻2兲t + 0.75 sin 共2␻1 − ␻2兲t
− 0.75 sin 共2␻2 + ␻1兲t + 0.75 sin 共2␻2 − ␻1兲t兴.
(5)
If we set the coefficient c3 ⳱ 0.30888 to produce 10%
THD at the single-tone input, and if we set the maximum
level of the two-tone signal equal to the amplitude of a
single tone, we obtain the level of harmonic components
listed in Table 2.
If someone compares the measurement results of har-
monic distortion produced by the two hypothetical suspen-
sions he might come to the conclusion that they perform
essentially similarly because their THD is equal to 10%,
the levels of their third harmonics are close (−21.7 dB
versus −22.3 dB), and the fifth harmonic produced by the
fifth-order suspension is small and not important. How-
ever, the larger number of intermodulation products pro-
duced by the fifth-order nonlinearity (which remained be-
(4) yond the scope of this particular harmonic measurement)
Table 1
Two Tones
Harmonics
1st 3rd 5th
−7.0 dB −37.8 dB −65.8 dB
J. Audio Eng. Soc., Vol. 52, No. 4, April 2004
PAPERS COMPARISON OF LOUDSPEAKER NONLINEAR DISTORTION MEASUREMENTS

might produce a different effect on the perceived sound affinity with a musical signal. Truly, the multitone stimu-
quality. Fig. 6 illustrates the output spectrum. The spec- lus is closer to a musical signal than the single-tone stimu-
trum of the distortion products is wider and the density of lus in the crest factor, the spectrum, and the probability
the spectrum is higher in the fifth-order suspension. It may density function. This example illustrates the dominance
cause higher perceptibility of nonlinear distortion in the of intermodulation products revealed by the multicompo-
fifth-order system because some distortion products might nent testing signals. The tendency of intermodulation
not be masked by a hearing system. products to dominate harmonics, illustrated here through
This simple example illustrates a situation when the the use of a multitone signal, can be extrapolated to a musical
measurement of only harmonics does not convey enough signal. More details on this subject can be found in [1].
information about the performance of even a simple static
nonlinear system. It also shows that a higher order static 3.2 THD and Harmonic Distortion
nonlinearity produces a larger number of intermodulation
By observing only harmonic distortion curves we might
products if excited by a similar input signal.
not be able to come to an accurate conclusion about the
Three conclusions follow.
entire nonlinear properties of a loudspeaker under test, and
1) The overall level of harmonics is typically lower than
we cannot predict how the distortion products generated in
the overall level of intermodulation products within the
a musical signal will be masked by the hearing system. In
same nonlinear system, and this difference is stronger in a
addition, harmonic distortion measurements may not re-
system impaired by a higher order nonlinearity.
veal some nonlinear effects at all. A typical example is the
2) A nonlinear system of a higher order being exposed
Doppler distortion in direct-radiating loudspeakers. This
to a complex signal produces more intermodulation prod-
distortion is not revealed by a single tone. At least two
ucts with a wider spectrum. This effect is not revealed by
tones are required.
an analysis of the harmonic distortion.
The “supremacy” of intermodulation distortion may
3) The wider spectrum of intermodulation products
lead to the straightforward but wrong conclusion that har-
might be more noticeable because some of the spectral
monic distortion is irrelevant in any application and may
components would not be masked by the hearing system.
be omitted in measurements of loudspeaker nonlinear dis-
Table 3 shows the reaction of second- and third-order
tortion. However, while not being able to characterize
static nonlinearity to different multitone stimuli. The num-
nonlinearity in its entirety and complexity, and link it to
ber of intermodulation products of static nonlinearity char-
the audibility of signal deterioration, the harmonic distor-
acterized by only second and third orders increases dra-
matically with the number of input testing tones compared
to the number of harmonics. It can also be observed that
the third-order nonlinearity produces the same number of
harmonic products (compared to the second order) but a
significantly larger amount of intermodulation products.
This tendency increases in higher order nonlinearities.
In this example the increase in the number of inter-
modulation products generated stems from the nature of
the testing signal (multitone) that was chosen for some

Table 2

Single Tone Two Tones

Harmonics Harmonics
1st 3rd 1st 3rd
−2.3 dB −22.3 dB −7.7 dB −40.3 dB Fig. 6. Spectrum of nonlinear reaction to two-tone input; third-
order approximation.

Table 3

Second-Order Nonlinearity Third-Order Nonlinearity Overall (Second and Third)

Number
of Initial Number of All Number of Number of All Number of Number of All Number of
Tones IM Products All Harmonics IM Products All Harmonics IM Products All Harmonics
1 0 1 0 1 0 2
2 2 2 4 2 6 4
3 6 3 16 3 22 6
4 12 4 40 4 52 8
5 20 5 80 5 100 10
10 90 10 660 10 750 20
15 210 15 2240 15 2450 30
20 380 20 5320 20 5700 40

J. Audio Eng. Soc., Vol. 52, No. 4, April 2004 339

VOISHVILLO ET AL.
tion curves, plotted separately as a function of frequency
and level of input signal, provide useful information about
loudspeaker under test. For example, a strong level of
high-order harmonics may be indicative of a rubbing voice
coil or the presence of nonlinear breakups of a compres-
sion driver’s metallic diaphragm and suspension. The re-
lationship between harmonics of even and odd orders tells
about the symmetry (or the lack of it) in loudspeaker dis-
placement-dependent parameters. The buildup of high-
order harmonics with an increase in input voltage may be
indicative of approaching the limit of a spider’s deflection.
When performing harmonic distortion measurements, one
should keep in mind that the harmonics will be accompa-
nied by an outweighing number of intermodulation prod-
ucts as soon as the testing tone is replaced by a musical
signal.
The THD test can be used legitimately in “passed–not
passed” production tests where similar types of loudspeak-
ers are tested. Certainly, THD gives an idea about audibly
noticeable nonlinear distortion if its level is high. A loud-
speaker having 50% THD in the midrange will hardly be
a source of mellifluous sound. It does not take some other
sophisticated analysis of nonlinearity or fine listening tests
to figure that out.
Fig. 7. SPL THD of 12-in (305-mm) woofer. Nonlinearity is
produced by Bl product, suspension stiffness, voice-coil induc-
tance, and flux modulation. Input voltage 10 V. (b) Same as (a),
but flux modulation omitted.
340
PAPERS
Fig. 7 illustrates the sensitivity of THD test results to
variations of the loudspeaker parameters. Fig. 7(a) corre-
sponds to input voltage of 10 V applied to a nonlinear
dynamic model of a 12-in (305-mm) woofer characterized
by a nonlinear Bl product, suspension stiffness, voice-coil
inductance, and flux modulation. The parameters of the
woofer are given in Appendix 1. The modeling was carried
out through numerical solution of a system of nonlinear
differential equations describing the behavior of an elec-
trodynamic loudspeaker (see Appendix 2). Fig. 7(b) shows
the THD curve corresponding to the same woofer, but the
flux modulation distortion is omitted. The difference in the
physical properties between the two models is reflected in
the difference in THD curves. It can be observed that the
flux modulation distortion affects the THD curve at high
frequencies. It is convenient to overlay THD curves cor-
responding to different input levels. Fig. 8 shows SPL
THD curves corresponding to an increase of the input
voltage from 10 to 40 V in 3-dB increments.
Figs. 9 and 10 show THD curves corresponding to two
8-in (203-mm) woofers having different motors. (One has
a long 12-mm coil and a 6-mm short gap, the other a short
6-mm coil and a long 12-mm gap.) The parameters of the
woofers are given in Appendix 1. The level of the input
signals corresponds to maximum voice-coil displacements
of 4 and 10 mm. The difference in the THD curves of the
small-level signal a is pronounced at low frequencies,
whereas the difference in the large-level signal b is pro-
nounced at frequencies above 80 Hz. Therefore THD gives
an idea of the difference in the objective performance of
two loudspeakers being compared.
3.3 High-Order Frequency Response Functions
(Frequency-Domain Volterra Kernels)
All the foregoing conclusions concerning the deficiency
of information about loudspeaker nonlinearity provided by
harmonic and THD measurements do not mean, however,
that a frequency response of traditionally measured indi-
vidual intermodulation products of the second or third or-
ders will always represent accurately the nonlinear prop-
erties of a loudspeaker in their entire complexity. The
following example will illustrate a situation when neither
Fig. 8. Increase in SPL THD of 12-in (305-mm) woofer corre-
sponding to increase in input voltage from 10 to 40 V in 3-dB
increments.
J. Audio Eng. Soc., Vol. 52, No. 4, April 2004
PAPERS COMPARISON OF LOUDSPEAKER NONLINEAR DISTORTION MEASUREMENTS

the second harmonic nor the second-order difference- order sum intermodulation product Pf2+f1. The diagonal line
frequency intermodulation products provides sufficient in- a, characterized by the equal in modulus but opposite in
formation about the distortion of a hypothetical loud- sign frequencies, describes the zero-order harmonic Pf1−f1,
speaker characterized by only a second-order nonlinearity. which is a constant displacement if the HFRF describes
Figs. 11 and 12 show second-order frequency responses the voice-coil excursion. The diagonal line characterized
(frequency-domain Volterra kernels of the second order)
of the same 8-in (203-mm) woofers having different mo-
tors. A typical HFRF is presented in the form of a three-
dimensional “mountain terrain” with two horizontal fre-
quency axes. Figs. 11(a) and 12(a) illustrate the three-
dimensional terrains of the second distortion products,
whereas Figs. 11(b) and 12(b) show the maps of this sur-
face. The vertical axis shows the level of all second-order
distortion frequency responses, including second harmon-
ics, the sum and difference intermodulation products at all
combinations of two frequencies, and the frequency-
dependent constant component (dc or zero harmonic) if it
is excited in a particular nonlinear system [see Figs. 11(a)
and 12(a)].
In this interpretation the Cartesian coordinates of a point
on this map corresponding to one negative and one posi-
tive frequency describe the second-order difference fre-
quency component Pf2−f1, whereas a point with the coor-
dinates belonging to both positive frequencies is a second-

Fig. 9. SPL THD corresponding to voice-coil maximum dis-

placement. Loudspeaker A (long coil, short gap). a—Xmax ⳱ 4
mm; b—Xmax ⳱ 10 mm.

Fig. 11. Sound pressure response of loudspeaker A (long coil,

Fig. 10. SPL THD corresponding to voice-coil maximum dis-
placement. Loudspeaker B (short coil, short gap). a—Xmax ⳱ 4
mm; b—Xmax ⳱ 10 mm.
short gap). (a) Second-order frequency-domain Volterra kernel.
Peak level of input signal corresponds to Xmax ⳱ 4 mm. (b)
Topological view of second-order SPL response. a—second har-
monic P2f; b—sum intermodulation product Pf2+f1; c—difference
intermodulation product Pf2−f1. Unique area of kernel is high-
lighted.

J. Audio Eng. Soc., Vol. 52, No. 4, April 2004 341

VOISHVILLO ET AL.
by the equal values of positive frequencies is a second
harmonic distortion Pf1+f1 (see Figs. 11 and 12). By com-
paring the second harmonic distortion cut with the entire
surface of the second-order distortion, one may clearly see
what a small share of all the information required to de-
Fig. 12. Sound pressure response of loudspeaker B (short coil,
long gap). (a) Second-order frequency-domain Volterra kernel.
Peak level of input signal corresponds to Xmax ⳱ 4 mm. (b)
Topological view of second-order SPL response. a—second har-
monic P2 f ; b—sum intermodulation product Pf2+f1; c—difference
intermodulation product Pf2−f1. Unique area of kernel is high-
lighted.
342
PAPERS
scribe entire second-order nonlinearity is represented by
the single second harmonic distortion response. To know
the dynamic reaction of the second-order nonlinearity we
should also take into account the three-dimensional sur-
face of the second-order phase response and perform a
twofold inverse Fourier transform, which will obtain the
second-order impulse response. Two-dimensional convo-
lution of the input signal with this two-dimensional pulse
response provides the dynamic reaction of the second-
order nonlinearity. It is not hard to imagine how far the
distortion signal may be from the results of harmonic or
THD measurement. Whether or not this dynamic distor-
tion signal is noticed by the hearing system depends on a
number of factors that should be considered in the context
of masking such as the level of the signal, its dynamics,
and the spectral contents.
The concept of Volterra expansion can be formally ex-
tended to higher orders. Unfortunately the third-order non-
linearity needs four-dimensional space for its description
(three frequency scales), which defies simple graphic rep-
resentation. One possible solution to plot the third-order
HFRFs is a cut through one of the three frequency scales
corresponding to the worst case of distortion, and using the
remaining two scales in the three-dimensional graph. Reed
and Hawksford used this approach in [17]. For the higher
orders of nonlinearity the situation becomes even more
desperate, and graphical representation is even less prac-
tical. To make matters worse, with increasing orders of
nonlinearity the volume of calculations required to de-
scribe a Volterra model increases tremendously, making a
practical application impossible. This “curse of dimen-
sionality” is clearly illustrated by an analysis of the ex-
pression for the output signal of a nonlinear system de-
scribed by the first three terms of a Volterra expansion,
t t t
y共t兲 = 兰h 共␶ 兲x共t − ␶兲 d␶ + 兰兰h 共␶ , ␶ 兲x共t − ␶ 兲
0
1 1
0 0
2 1 2 1
t t t
× x共t − ␶2兲 d␶1 d␶2 + 兰兰兰h 共␶ , ␶ , ␶ 兲x共t − ␶ 兲
0 0 0
3 1 2 3 1
× x共t − ␶2兲x共t − ␶3兲 d␶1 d␶2 d␶3. (6)
Here h2(␶1, ␶2) is a second-order kernel, or a two-
dimensional impulse response, which depends on the two
time arguments ␶1 and ␶2. Correspondingly, h3(␶1, ␶2, ␶3)
is a third-order kernel, depending on three time arguments.
If, for example, the analysis of the first-order kernel
(linear impulse response) is carried out on one thousand
samples, the second-order kernel (second-order impulse
response) requires one-quarter million samples to be ana-
lyzed with analogous resolution. The analysis of the third-
order kernel with the same resolution amounts to 375 mil-
lion samples, whereas the fourth-order kernel requires 250
billion samples to keep the same accuracy. In this estima-
tion the property of Volterra kernel symmetry was used
and the redundant parts of kernels were omitted.
Unique areas of the second-order kernels presented are
highlighted in Figs. 11(b) and 12(b). One quarter of all
possible permutations of the samples of the second-order
system is required to describe the system. Three-eighths of
J. Audio Eng. Soc., Vol. 52, No. 4, April 2004
PAPERS
all possible permutations of the third-order system’s re-
sponse samples is enough to describe it completely, and
one-fourth of all permutations of the fourth-order samples
[14]. Still, the number of samples to be analyzed increases
enormously with an increase in the order of Volterra
expansion.
The calculation of HFRFs from multidimensional pulse
responses needs n-fold Fourier transforms. The volume of
calculations also increases significantly with an increase in
the order of nonlinearity,
⬁ lation products. However, it is known from the theory of
H1共i␻1兲 = 兰h 共␶ 兲e
−⬁
1 1
−i共␻1␶1兲
d␶1
⬁
H2共i␻1, i␻2兲 = 兰h 共␶ , ␶ 兲e
−⬁
2 1 2
−i共␻1␶1+␻2␶2兲
d␶1 d␶2
⬁ It is not practical to plot 16 different frequency re-
H3共i␻1, i␻2, i␻3兲 = 兰h 共␶ , ␶ , ␶ 兲e
3 1 2 3
−i共␻1␶1+␻2␶2+␻3␶3兲
−⬁
d␶1 d␶2 d␶3
⭈ intermodulation products of the second and third orders
⭈
⭈ gives limited information about the third-order nonlinear-
⬁
兰h 共␶ , . . . , ␶ 兲e −i共␻1␶1+⭈⭈⭈+␻n␶n兲
Hn共i␻1, . . . , i␻n兲 = n 1 n methods supply limited information as well.
−⬁
d␶1 . . . d␶n.
In addition, the Volterra series expansion has a fundamen-
tal constraint stemming from the assumption that there is
no energy exchange between nonlinear products of differ-
ent orders. This constraint confines the application of
Volterra series expansions to only weakly nonlinear sys-
tems. Attempting to use Volterra expansions for a nonlin-
ear system with a strong nonlinearity causes divergence of
the Volterra series.
This simple example illustrates why Volterra expan-
sions of orders higher than three are practically never used.
It precludes Volterra expansion from handling strong and
high-order nonlinearities. The measurement of Volterra
HFRFs can be carried out using special signals, such as
maximum-length sequences (MLS) [27], multitone stimuli
[28], and Gaussian noise [29]. There are methods provid-
ing a direct calculation of HFRFs from NARMAX output
data [30]. Straightforward methods operating with a varia-
tion of two or three sinusoidal signals are not practical
because of the measurement time burden.
3.4 Two-Tone Intermodulation Frequency
Responses
Measuring the frequency responses of intermodulation
products by using a two-tone signal has nearly as old a
history as measuring harmonic distortion and THD. Two
methods have been used predominantly in the audio in-
dustry. One, proposed by Hilliard, uses one fixed low-
frequency tone and one sweeping tone. The method was
adopted by SMPTE [19]. It is often called the modulation
method. The second method, using two sweeping tones
and keeping the frequency difference between them con-
J. Audio Eng. Soc., Vol. 52, No. 4, April 2004
COMPARISON OF LOUDSPEAKER NONLINEAR DISTORTION MEASUREMENTS
stant, was proposed by Scott [20]. It is called difference
frequency or CCIF method. If a two-tone signal is applied
to a second-order nonlinear system, it generates the fol-
lowing distortion products: dc component, two different
intermodulation spectral components (sum and difference
products), and two second harmonics. Meanwhile the non-
linear reaction of the third-order nonlinearity to the same
two-tone signal will consist of two third-order harmonics,
two spectral components having the same frequencies as
the initial tones but lower amplitudes, and four intermodu-
nonlinear systems that full description of the third-order
nonlinearity formally needs at least a three-tone signal
[14]. This increases the number of third-order harmonics
to three, and the number of spectral components having
input signal frequencies to three as well. The number of
intermodulation products goes as high as 16.
sponses of intermodulation products. Traditionally only
the products Pf2±2f1 are analyzed, omitting components of
(7)
the type Pf3±f2±f1, which are also generated by the third-
order nonlinearity. Hence the measurement of individual
ity if a two-tone signal is used. With regard to higher order
nonlinearity, the standardized two-tone intermodulation
Plotting all four “conventional” intermodulation curves
(Pf2±f1 and Pf2±2f1) on a single graph still produces a picture
that is difficult to comprehend and interpret. An integrated
criterion in the form of total intermodulation distortion
(TIMD) is a simplifying solution, leading to fewer fre-
quency responses of intermodulation distortion. For ex-
ample, standard IEC 60268-5 [22] determines: “The
modulation distortion (MD) of the nth order shall be speci-
fied as the ratio of the arithmetic sum of the r.m.s. values
of the sound pressures due to distortion components at
frequencies f2 ± (n − 1)f1 to the r.m.s. value of the sound
pressure Pf2 due to the signal f2.” The total intermodulation
coefficient of the second order according to IEC 60268-5 is
Pf2−f1 + Pf2+f1
d2 = × 100%. (8)
Pf2
The total intermodulation coefficient of the third order is
Pf2−2f1 + Pf2+2f1
d3 = × 100%. (9)
Pf2
The frequencies f1 and f2 satisfy the condition f2 >> f1, and
the ratio of the amplitudes of the input signal is specified
by the user. The standard gives no recommendation re-
garding the measurement of intermodulation and harmonic
products having orders higher than three. This omission
was probably due to practical concerns. Without calling
into question the validity of the standard’s recommenda-
tions, the authors do not exclude situations when measur-
ing higher order harmonic and intermodulation products
might be useful in the assessment of audio equipment
performance.
343
VOISHVILLO ET AL. PAPERS

As was mentioned in Section 2, higher order nonlinear tone intermodulation distortion (TIMD) is specified by the
products may be more detrimental to the sound quality authors as

冑兺
compared to the lower order products. This effect was
N
recognized a long time ago (see, for example, [31]), and
modern research confirms it [1], [32]. The importance of p2i 共 f 兲
i=1
higher order distortion products is somewhat twofold. dTIMD共f 兲 = × 100% (10)
First, high-order nonlinearity produces a very large num- Pf
ber of intermodulation products, whose number and en- where
ergy increase dramatically with an increasing input signal
level. Second, higher order products are usually spread p1共 f 兲 = Pf2+f1, p2共 f 兲 = Pf2−f1, p3共 f 兲 = Pf2+2f1,
over a wide frequency range, which results in weaker psy- p4共 f 兲 = Pf2−2f1, p5共 f 兲 = Pf2+3f1, p6共 f 兲 = Pf2−3f1,
choacoustic masking of these distortion products [1]. In p7共 f 兲 = Pf2+4f1, p8共 f 兲 = Pf2−4f1, . . .
the wake of it, the authors developed an alternative way to pn−1共 f 兲 = Pf2+mf1, pn = Pf2−mf1
formulate two-tone intermodulation distortion characteris-
tics. Figs. 13 and 14 show the two-tone intermodulation are the amplitudes of the intermodulation products, Pf ⳱
distortion curves of the same two 8-in (203-mm) woofers. Pf1 ⳱ Pf2 is the amplitude of either one of the fundamental
The distortion curves correspond to similar 10-mm maxi- tones, f1 is the fixed low-frequency tone, and f2 is the
mum voice-coil displacement. The difference between higher frequency sweeping tone.
these intermodulation distortion curves of the two loud- In this approach the excitation signal consists of two
speakers is significantly more pronounced when compared tones having equal amplitude. One of these two tones is
with the THD curves of the same woofers. The intermodu- swept across the frequency range. The level of distortion,
lation distortion curves presented in Figs. 13 and 14 are calculated according to Eq. (10), is plotted at the fre-
calculated differently from traditional intermodulation co- quency of the sweeping tone. The authors attempted to
efficients recommended by existing standards. The two- extend the recommendations of IEC 60268-5 [22] to the

Fig. 13. Two-tone intermodulation distortion (IEC 60268-5);
loudspeaker A (long coil, short gap). a—(Pf2−f1 + Pf2+f1)/Pf1;
b—(Pf2−2f1 + Pf2+2f1)/Pf1. (a) Voice-coil maximum displacement
Xmax ⳱ 4 mm. (b) Voice-coil maximum displacement Xmax ⳱
10 mm.
Fig. 14. Two-tone intermodulation distortion (IEC 60268-5);
loudspeaker B (short coil, long gap). a—(Pf2−f1 + Pf2+f1)/Pf1;
b—(Pf2−2f1 + Pf2+2f1)/Pf1. (a) Voice-coil maximum displacement
Xmax ⳱ 4 mm. (b) Voice-coil maximum displacement Xmax ⳱
10 mm.

344 J. Audio Eng. Soc., Vol. 52, No. 4, April 2004

PAPERS COMPARISON OF LOUDSPEAKER NONLINEAR DISTORTION MEASUREMENTS

measurement of two-tone intermodulation distortion in the total harmonic and intermodulation distortion is extended
form of the intermodulation products having frequencies to a larger number of input tones. The ten-tone total har-
(f2 ± f1) and (f2 ± 2f1). The latter are the products of the monic distortion coefficient takes into account the har-
interaction between the first and second harmonics; the monics of all input tones. The level of this distortion,
first tone with the first harmonic of the second tone. The similar to the TTIMD, is related to one of the fundamental
authors merely extended the set of harmonics of the first tones. This simplifies the comparison of the distortions
signal to higher orders (3f1, 4f1), which led to intermodu- evaluated by these two criteria.
lation terms of the kind (f2 ± 3f1), (f2 ± 4f1). The inter-
modulation terms corresponding to the interaction be- 3.5 Multitone Stimulus
tween the harmonics of the first tone and the harmonics of
The possible circumvention of the partial “blindness” of
the second tone, such as (2f2 ± 2f1), (3f2 ± 2f1), (2f2 ± 3f1),
the conventional two-tone intermodulation tests is not to
(3f2 ± 3f1), etc., were omitted. The authors do not claim the
plot continuous frequency responses of the corresponding
ultimate validity of this approach. Including all the inter-
intermodulation products, but rather to show the full dis-
modulation products produced by two-tone excitation
crete spectra of all nonlinear products corresponding to
would probably provide more accurate results.
particular frequencies and levels of the two test tones. By
An alternative method to measure intermodulation dis-
extending this idea to a larger number of excitation tones,
tortion has been used by Keele [23], [24]. His test signal,
we naturally arrive at the concept of the multitone signal.
consisting of two tones, 40 Hz and 400 Hz, of equal am-
Indeed, if we obtain and graph the spectrum of a nonlinear
plitude, is applied to a loudspeaker, the input level is in-
reaction to the two-tone signal, which, as it has been
creased, and the intermodulation is measured and plotted
shown, gives limited information even about the third-
as a function of the input level. Such a test is a simple way
order nonlinearity, let alone the higher orders, why not use
to evaluate the intermodulation of the midrange output of
as many tones as it takes to detect all conceivable higher
a loudspeaker by a simultaneous bass signal.
In the current work Keele’s general approach to mea-
suring intermodulation distortion versus input level was
simulated using two different criteria. The first, dTTIMD,
includes all N measurable output intermodulation prod-
ucts; the second, dTTHD, takes into account only M har-
monic distortion products produced by two primary tones.
Here TTIMD stands for two-tone total intermodulation
distortion and TTHD designates two-tone total harmonic
distortion

dTTIMD =
冑兺 N

i=1
P共i兲IM
2

× 100% (11)
Pf

dTTHD =
冑兺 M

k=1
P共k兲H2
× 100% (12)
Pf

where P(i)IM is the amplitude of an intermodulation prod-

uct corresponding to the ith frequency, P(k)H is the am-
plitude of the harmonic corresponding to the kth fre-
quency, and N and M are the number of intermodulation
and harmonic products, respectively. To avoid the over-
lapping of the fundamentals and the distortion products,
the frequencies of the primary tones were chosen as f1 ⳱
fs and f2 ⳱ 5.5fs with frequency fs being the resonance
frequency of the loudspeaker.
Figure 15 shows the graphs of two-tone total intermodu-
lation and harmonic distortion of two 8-in (203-mm)
woofers as a function of the input level. The frequencies of
the tones are 57 and 313.5 Hz. The harmonic distortion of Fig. 15. Two-tone total nonlinear distortion as a function of input
the loudspeaker with the long voice coil a prevails at the voltage. a—intermodulation products; b—harmonic products. (a)
Loudspeaker A (long coil, short gap.). f1 ⳱ fs ⳱ 57 Hz; f2 ⳱
lower level of the input signal. However, this effect is not 5.5 ⭈ f1 ⳱ 313.5 Hz. Umax ⳱ 29 V corresponds to Xmax ⳱ 10 mm.
observed when the two-tone signal is replaced by a ten- (b) Loudspeaker B (short coil, long gap). f1 ⳱ fs ⳱ 62 Hz; f2 ⳱
tone input signal (Fig. 16). Here the concept of two-tone 5.5 ⭈ f1 ⳱ 341 Hz. Umax ⳱ 31 V corresponds to Xmax ⳱ 10 mm.

J. Audio Eng. Soc., Vol. 52, No. 4, April 2004 345

VOISHVILLO ET AL.
order intermodulation products, cover the entire frequency
range of measurement, and have a signal statistically much
closer to the real musical signal than a single-tone or two-
tone signal?
This idea can be extended to a tone in every FFT fre-
quency bin. This abundance of tones turns the multitone
stimulus into a noiselike signal. Truly, noise signals are
used widely in the identification and analysis of nonlinear
systems as, for example, in the measurement of the coher-
ence function. However, once the FFT is applied to the
output signal of a nonlinear system excited by such noise-
like signals, all individual distortion spectral components
are obscured by the fundamental tones and become invis-
ible on a graph. Meanwhile the multitone signal, produc-
ing a “sparse” and discrete spectrum at the output of a
nonlinear system, makes the majority of distortion prod-
ucts visible on a graph. At the same time the multitone
signal is rather close to noise and musical signals in the
probability density function, bandwidth, and crest factor.
The multitone stimulus fills the gap between the noise-
based methods of nonlinear identification and measure-
ments, and the traditional standardized methods using one
or two stationary or swept (stepped) tones.
Fig. 16. Ten-tone total nonlinear distortion as a function of input
voltage. a—intermodulation products; b—harmonic products.
Logarithmic frequency distribution in frequency range fs to 5.5 fs.
(a) Loudspeaker A (long coil, short gap). Umax ⳱ 7.3 V corre-
sponds to Xmax ⳱ 10 mm. (b) Loudspeaker B (short coil, long
gap). Umax ⳱ 7.8 V corresponds to Xmax ⳱ 10 mm.
346
PAPERS
Usually the multitone signal is generated according to
the simple rule
N
兺A sin 共␻ t + ␸ 兲.
x共t兲 = i i i (13)
i=1
A strong advantage of the multitone stimulus is a short
measurement time and the ability to reveal simultaneously
a set of visible harmonic and intermodulation products. In
this capacity the multitone signal is beyond competition
with other signals. Multitone testing handles high-order
nonlinearity, and its use is not hampered by the existence
of such effects as hard limiting, hysteresis, and dead zone.
Also, the multitone stimulus can be used in applications
where the loudspeaker short-term performance must be
evaluated, such as the maximum SPL. Comparing a mul-
titone burst with a tone burst, the advantages of the former
become obvious. After the time-domain reaction of the
loudspeaker to the tone burst of a particular frequency has
been received and preprocessed to skip the transients and
then put through the Fourier transform, only harmonic
distortion and THD become available. The distortion (har-
monic or THD) corresponds to only a single excitation
frequency. To cover the whole frequency range of interest,
these measurements have to be repeated at different fre-
quencies. Taking into account the number of measure-
ments needed to cover the entire frequency range with a
decent resolution, the overall measurement time may be
significant. Meanwhile, by applying the multitone burst,
only one measurement is needed and, in addition, the mul-
titone signal obtains more information about the nonlin-
earity in a loudspeaker under test. Furthermore, a multi-
tone burst’s crest factor can be “tuned” by adjusting the
phases of individual spectral components.
However, the interpretation of a nonlinear reaction to a
multitone stimulus may be arduous if the number of gen-
erated nonlinear products of different orders is substantial.
A multitone stimulus gives such “abundant” spectral in-
formation about nonlinearities that it is difficult to com-
prehend at first sight. (Truly, a second look at the reaction
to the multitone stimulus may not be helpful either when
one has to analyze hundreds if not thousands of distortion
spectral components.) In addition, an engineer has no in-
formation on how a particular pattern of nonlinear reac-
tions to multitone stimuli is related to the perceived sound
quality. Moreover, the spectrum of reactions to multitone
stimuli is not convenient to overlay and compare, espe-
cially if the responses to several input levels are to be
observed. There are several possible ways to overcome
this impediment. One is to distinguish the products of
different orders by postprocessing and to plot them either
separately or in different colors on the same graph. An-
other solution is to plot the averaged value of all distor-
tions located between two adjacent tones, as it is done in
[10] and in the FASTTEST multitone measurement [8].
This approach permits a simple graphical representation of
nonlinear distortions at different levels of input signal.
Distinguishing different intermodulation and harmonic
products of different orders is comparatively easy when
the number of initial tones is reasonably low (less than ten,
J. Audio Eng. Soc., Vol. 52, No. 4, April 2004
PAPERS COMPARISON OF LOUDSPEAKER NONLINEAR DISTORTION MEASUREMENTS

for example). With an increasing number of input tones
overlapping of different frequencies occurs, and the prob-
lem of separation becomes much more difficult, but not a
theoretically impossible task. The separation can be car-
ried out by a discrete progressive increase of the input
level accompanied by an analysis of the rate of increase of
each distortion product in a particular frequency bin in
such a way that the evaluated spectral components (in-
cluding possible overlapped ones) measured at different
levels of input signal form a so-called polynomial Van-
dermonde matrix [33]. Corresponding mathematical ma-
nipulations with this matrix, which remain beyond the
scope of this work, provide a separation of the overlapped
spectral components and make it possible to evaluate the
level and phase of each, disregarding the fact that they
overlap. This approach is described, for example, in [28],
where a multitone signal is used in the identification of
weakly nonlinear systems and the measurement of Volt-
erra kernels.
An alternative method to represent the results of multi-
tone testing is the averaging of distortion products in a
“sliding window.” The spectral components are averaged
in a window (such an rectangular or Hanning), and the
averaged value of the distortion products is plotted at the
frequency corresponding to the center of the window [34].
Afterward the window is shifted one frequency bin “up”
and the process is repeated. Ultimately it provides a con-
tinuous frequency response of the distortion products,
which encapsulates all harmonics and a variety of inter-
modulation products generated by a particular loudspeaker
at a particular level of multitone stimuli, and a particular
distribution of primary tones. It has been dubbed the mul-
titone total nonlinear distortion (MTND). One of the pos-
sible ways to calculate MTND, where the Hanning win-
dow is used, is presented in the expression
dMTND共 fi兲 =
This expression for the MTND characteristic uses a rect-
angular window and the weighting coefficient 1/K, where
K is the number of distortion products in the rectangular
window. This way to formulate dMTND(f) gives values that
are too low if the number of distortion products in a cur-
rent window is significant but the level of them is not high.
This statement is purely empirical and bears no relation-
ship to subjective sensations. An alternative way might be
to omit the weighting coefficient 1/K entirely. In this case,
however, the level of dMTND(f) may become dispropor-
tionally high if the number of distortion products corre-
sponding to a particular position of the rectangular win-
dow is high, even if their amplitude is low. These practical
considerations mean that when the level of an MTND
curve is much lower or much higher than the level of
distortion spectral components, the graph of distortion
looks unnatural. This is merely the authors’ subjective
point of view derived from numerous modeling and mea-
surement experiments.
Fig. 17 shows the spectrum of the SPL reaction to
the input multitone stimulus of the same two 8-in (203-
mm) woofers. The solid curves correspond to MTND
calculated according to Eq. (15). Fig. 18 shows the
reaction to multitone stimuli of the 12-in (305-mm)
woofer with and without flux modulation distortion. The

20 log 冉冑 兺 再 冋 冉 冊 册 冎 冒 冊
KⲐ2

k=i−K Ⲑ 2
Dk cos
␲| fi − fk|
⌬f
+1
1
2
2

p0

共dB SPL兲 (14)

where ⌬f is the width of the frequency window, fi is the
window center frequency, Dk is the amplitude of a sound
pressure distortion product (Pa) at the frequency fk , and K
is the number of spectral components; p0 ⳱ 2 × 10−5 Pa.
The frequency window consists of K frequency bins.
The window is essentially a weighting function that has a
maximum at the frequency fi corresponding to the center
of the window. This way of formulating multitone distor-
tion was chosen experimentally. The resulting frequency-
dependent function dMTNDf) looks like an envelope of the
distortion products.
There are other possible ways to express multitone total
nonlinear distortion. For example, dMTND(f) can be ex-
pressed as

dMTND共 f1兲 = 20 log 冉冑 1

兺
KⲐ2

D2
Kk=i−K Ⲑ 2 k 冒冊
p0 共dB SPL兲.
Fig. 17. Sound pressure reaction to multitone stimulus. Peak
level of input signal corresponds to Xmax ⳱ 10 mm. (a) Loud-
speaker A (long coil, short gap). (b) Loudspeaker B (short coil,
(15) long gap).

J. Audio Eng. Soc., Vol. 52, No. 4, April 2004 347

VOISHVILLO ET AL.
solid curves correspond to MTND calculated according to
Eq. (14).
The presentation of the reaction of a device impaired by
nonlinearity to multitone stimuli in the form of an aver-
aged curve (MTND) makes it easy to overlay different
curves belonging, for example, to different levels of input
signals or to different loudspeakers. Fig. 19 shows two over-
laid MTND curves, indicating that the flux modulation pro-
duces distortion in the upper part of the frequency range.
The frequency response of the MTND curve can be
expressed in dB SPL [Eqs. (14) and (15)] as well as in the
percentage of the fundamental frequency response,
dMTND共 fi兲
冉冑 兺 再 冋 冉 冊 册 冎 冒 冊
KⲐ2 2 standardized.
␲| fi − fk| 1 There are many methods of forming the frequency dis-
= Dk cos +1 A共 fi兲
k=i−K Ⲑ 2 ⌬f 2 tribution of multitone fundamentals. The major goal of
× 100共%兲 (16)
where A(fi) is the amplitude of the frequency response of
a loudspeaker at the frequency fi.
Fig. 20 shows MTND responses of the 12-in (305-mm)
woofer calculated according to Eq. (16) and at different
levels of the input signal in 3-dB increments.
Fig. 18. Multitone reaction and MTND (solid curve) of 12-in
(305-mm) woofer. (a) Nonlinearity is produced by Bl product,
suspension stiffness, voice-coil inductance, and flux modulation.
Input voltage U ⳱ 3.3 V; Xmax ⳱ 2.5 mm. (b) Same as (a), but
flux modulation not taken into account.
348
PAPERS
There is a current impediment to the widespread use of
multitone stimuli for measuring nonlinearity in loudspeak-
ers. This is the ambiguity of the nonlinear reaction of a
particular device to a multitone signal. A different distri-
bution and a different number of tones produce different
reactions. In theory all these responses belong to the same
multidimensional space of nonlinear reactions; however,
for an observer these responses look different. This com-
plicates the comparison of responses measured using
different distributions and number of tones. So the cur-
rent disadvantage of using multitone stimuli is the lack
of a common agreement regarding the number of tones,
their distribution, and the initial phases. To avoid this
problem the number and distribution of tones should be
some of the frequency distributions of primary tones (dif-
ferent from the evenly distributed tones on a logarithmic
frequency scale) is to minimize the overlapping of primary
tones and distortion components [7], [12].
As was mentioned, the separation becomes increasingly
difficult with an increasing order of the distortion products
due to the effect of overlapping. The separation can be
Fig. 19. SPL MTND corresponding to 12-in (305-mm) woofer. U
⳱ 3.3 V; Xmax ⳱ 2.5 mm. a—flux modulation taken into ac-
count; b—flux modulation omitted.
Fig. 20. MTND of 12-in (305-mm) woofer corresponding to
increasing level of input voltage. Umin ⳱ 3.3 V; voltage incre-
ments 3 dB.
J. Audio Eng. Soc., Vol. 52, No. 4, April 2004
PAPERS
handled through the use of the polynomial Vandermonde
matrix [28], [33], which is not a trivial procedure. The
separation of low-order and high-order distortion compo-
nents can be performed easily by, for example, a two-tone
signal. This is important for the detection of loudspeaker
defects (rub and buzz) separate from regular motor and
suspension nonlinearities. The simplicity of the distorted
two-tone signal allows one to “understand” the relation-
ship between some of the loudspeaker nonlinear param-
eters (causes) and nonlinear distortion (symptoms).
3.6 Coherence and Incoherence Functions
The next method that deserves discussion is the mea-
surement of the coherence function that characterizes the
degree of linear relationship between input and output as a
function of frequency. By definition, the coherence func-
tion is expressed as the ratio of the square of the cross
spectrum (between input and output) to the product of the
autospectra of input and output [35],
|Gxy共 fi兲|2
␥2共 fi兲 =
Gxx共 fi兲Gyy共 fi兲
where Gxx(fi) is the autospectrum of the input signal x(t) at
the frequency fi, Gyy(fi) is the autospectrum of the output
signal y(t), and Gxy(fi) is the cross spectrum of the input
signal x(t) and the output signal y(t).
The functions Gxx(fi), Gyy(fi), and Gxy(fi) are calculated
as follows:
N
兺
1 V rms set for the measurement of THD SPL (see Fig. 8),
Gxx共 fi兲 = E 关X共 fi兲 X*共 fi兲兴 = lim Xn共 fi兲 X*n 共 fi兲
N→⬁ N n=1
N
兺
1
Gyy共 fi兲 = E 关Y共 fi兲 Y*共 fi兲兴 = lim Yn共 fi兲 Y*n 共 fi兲
N→⬁ N n=1
N three methods, they produce different data, all related to
兺
1
Gxy共 fi兲 = E 关X共 fi兲 Y*共 fi兲兴 = lim Xn共 fi兲 Y*n 共 fi兲
N→⬁ N n=1
where * denotes complex conjugation, E indicates aver-
aging, and X( f ) and Y( f ) are the complex spectra of the
input and output signals x(t) and y(t), respectively,
兰
⬁
X共 f 兲 = F兵x共t兲其 = x共t兲e−j2␲ft dt
−⬁
Y共 f 兲 = F兵y共t兲其 = 兰
⬁
y共t兲e−j2␲ft dt
−⬁
with F{⭈} being the Fourier transform.
In a strictly linear noiseless system the coherence func-
tion ␥2(f) equals unity at all frequencies. To the contrary,
if the input x(t) and the output y(t) have no relation to each
other, the coherence function is zero. If the coherence
function has a value between 0 and 1, the system under test
may either be impaired by the nonlinear distortion or the
noise, or both, or the output y(t) depends on some other
input processes along with x(t). Therefore the coherence
function may be used as a measure of nonlinearity in a
device under test.
Historically use of the coherence function has never
been immensely popular in loudspeaker testing. Mean-
J. Audio Eng. Soc., Vol. 52, No. 4, April 2004
COMPARISON OF LOUDSPEAKER NONLINEAR DISTORTION MEASUREMENTS
while this method has found an application in the assess-
ment of nonlinearity in hearing aids [3], [4]. The coher-
ence function gives an integral lumped measure of the
nonlinearity in a device under test, but it also takes into
account noise if it is presented in a device under test. In
loudspeaker testing the presence of noise does not seem to
be an impeding factor. The attractive feature of the coher-
ence function is its simple graphic representation. Plotting
the set of coherence functions corresponding to different
input levels is another nice option.
It seems to be more convenient to present the coherence
function in the following manner and call it the incoher-
ence function,
I共 f 兲 = 公1 − ␥2共 f 兲 × 100共%兲. (23)
Expressed in percent, the incoherence function I(f) is in-
tuitively close to the concept of nonlinear distortion. Zero
incoherence indicates the absence of nonlinear distortion
and noise. There is a seeming similarity between the in-
coherence function and THD. However, there is a princi-
(17) pal difference between these two characteristics. THD
takes only harmonics into account, whereas the incoher-
ence function is sensitive to the overall nonlinear contami-
nation of the output signal and noise.
Fig. 21 shows the incoherence function of the 12-in
(305-mm) woofer corresponding to different levels of the
input signal. The initial level of the noise signal was 0.6 V
rms. This level produced a voice-coil peak displacement of
2.5 mm. The same peak displacement corresponded to 10
(18) and to 3.3 V rms for the multitone measurement (see Fig.
20). This difference in initial rms levels is attributed to
different crest factors of these signals. The incoherence
(19) function, THD, and MTND each show a different pattern
of nonlinear distortion. Due to the different nature of these
(20) the same particular nonlinearity. This example demon-
strates the complexity of assessment of nonlinear effects
and the nontrivial reactions of a nonlinear system to dif-
ferent testing signals. Fig. 22 shows the difference be-
tween the incoherence functions of two 8-in (203-mm)
woofers corresponding to voice-coil displacement of 4 and
(21)
(22)
Fig. 21. Sound pressure incoherence function of 12-in (305-mm)
woofer corresponding to increasing level of input voltage. Umin
⳱ 0.6 V; voltage increments 3 dB.
349
VOISHVILLO ET AL.
10 mm. An increase in the nonlinear distortion corre-
sponding to the increasing input signal can be observed.
The incoherence function was calculated by means of
the noise signal generated as a multitone signal with 4096
frequency components of equal amplitude and the random
distribution of phases. The sampling frequency was 7680
Hz. The crest factor of this signal is 5.9. To adjust the
properties of this noise signal for the numerical integration
of the system of nonlinear differential equations governing
the operation of a loudspeaker, an adaptive algorithm was
used to provide the initial zero value of the testing signal.
In the given examples the incoherence function resulted
from 1000 averages, which would correspond to approxi-
mately 500 seconds of testing time. During this time the
warming of the voice coil would change the behavior of
the loudspeaker significantly if the driver were operated at
high amplitudes. This is a drawback of this technique.
4 CONCLUSION
Due to the complex nature of loudspeaker nonlinearity
and the intricacy of the human auditory system’s reaction
to musical signals contaminated with nonlinear distortion
products, there are no undisputedly credible and com-
monly recognized thresholds of traditional nonlinear dis-
Fig. 22. Sound pressure incoherence function. a—peak level of
input signal corresponds to Xmax ⳱ 4 mm; b—peak level of input
signal corresponds to Xmax ⳱ 10 mm. (a) Loudspeaker A (long
coil, short gap). (b) Loudspeaker B (short coil, long gap).
350
PAPERS
tortion measures related to the perceived sound quality.
Since the dynamic reaction of a complex nonlinear system
such as a loudspeaker cannot be extrapolated from its re-
action to simple testing signals, such as a sweeping tone,
the thresholds expressed in terms of the loudspeaker reac-
tion to these signals (THD, harmonics, and two-tone in-
termodulation distortion) may not be valid.
The requirements for an optimal method of measuring
nonlinear distortion in loudspeakers were formulated. The
optimal method to measure the nonlinearity in loudspeak-
ers must be informative, that is, it must obtain enough
objective information about the nonlinearity of different
orders. The plotted measurement results must have a clear
interpretation and be readily comprehensible. The mea-
surement data must be supported psychoacoustically,
meaning that there should exist an unambiguous relation-
ship between the results presented and the expected sound
quality.
In nonlinear systems such as a loudspeaker, the inter-
modulation distortion outweighs the harmonic distortion if
a musical signal is reproduced. Harmonics may not give a
quantitative measure of the nonlinear distortion in a loud-
speaker, especially in the context of nonlinear distortion
audibility. Nevertheless, the harmonic distortion measure-
ment provides valuable information, illustrating, for ex-
ample, the dominance of the nonlinearity of certain orders.
A wide spectrum of harmonics and a strong level of high-
order harmonics may be indicative of a loudspeaker mal-
function such as a rubbing voice coil.
It has been demonstrated that high orders of static non-
linearity are characterized by a significant difference be-
tween the harmonic and intermodulation products that out-
number the harmonics and outweigh them in power. It has
also been demonstrated that a high-order nonlinearity pro-
duces intermodulation and harmonic products of its “own”
order, and of lower orders as well. The latter might have
higher levels. Drawing the conclusion that a certain high-
order nonlinearity is not essential because it produces a
low level of its “own” harmonics may lead to wrong
results.
THD does not seem to be a good measure of psycho-
acoustically meaningful distortion in loudspeakers. Not
distinguishing different orders of harmonics, the THD fre-
quency response may lead to the wrong conclusions about
the performance of a loudspeaker. Similar levels of THD
may correspond to very different distributions of harmon-
ics of different orders. This difference, invisible to THD,
may correspond to a strong diversity in intermodulation
products and correspondingly significant differences in
sound quality. However, THD can be legitimately used in
testing where similar types of loudspeaker are compared
(for example, in production testing).
Multitone testing possesses a number of advantages
compared to other methods. It is fast and gives a detailed
graphical representation of the distortion products. When a
large number of input tones are applied to a loudspeaker,
the spectrum of the output signal becomes very rich with
intermodulation products (harmonic products have only a
minuscule share of these spectral components). A visual
examination of such a spectrum, though, may be difficult.
J. Audio Eng. Soc., Vol. 52, No. 4, April 2004
PAPERS COMPARISON OF LOUDSPEAKER NONLINEAR DISTORTION MEASUREMENTS
To circumvent this problem, the spectral components of
different orders can be plotted separately. The separation
of products of different orders needs postprocessing.
Another option to simplify the visual interpretation of
measurement results is to plot the average level of dis-
tortion confined within adjacent tones or to plot the level
of the distortion products averaged in a sweeping fre-
quency window. The comparatively high crest factor of
multitone signals may give a “pessimistic” evaluation of
the distortion level, registering the low-probability high-
level peaks that may not be psychoacoustically relevant
when a testing signal is replaced by a musical one. More
research is required to put a reliable bridge between a
loudspeaker’s response to multitone stimuli and the sound
quality of the loudspeaker. The results of recently pub-
lished psychoacoustical research of a correlation between
the responses to multitone stimuli and the audibility of
distortion [26] imply that such a goal might possibly be
reached.
The incoherence functions of two 8-in (203-mm) woof-
ers were modeled at two different levels of the input noise
signal. In addition, the incoherence function of a 12-in
(305-mm) woofer was modeled for different levels of in-
put signal. The incoherence function detected the differ-
ence in performance of the two motors, showing an in-
crease in the overall nonlinear distortion for a loudspeaker
having stronger voice-coil inductance modulation and
stronger dependence of the Bl product on the voice-coil
displacement.
There is a significant difference between THD, incoher-
ence function, and reaction to multitone stimuli. All three
methods provide an “integral” assessment of nonlinear
distortion. However, the information conveyed by these
methods is principally different. THD characterizes only
harmonic distortion, omitting the intermodulation prod-
ucts, which significantly outweigh harmonics in a dis-
torted musical signal. The incoherence function expressed
in percent may be interpreted as a measure of the “lack of
similarity” between the reference and the output signal.
Contrary to THD, the incoherence function takes into ac-
count all nonlinear transformations of the signal as well as
the influence of noise. However, this function does not
distinguish the products of different orders, giving a
“lumped” integral measure. The multitone stimulus pro-
vides information about harmonic and intermodulation
products of various orders, but does it in a more diversified
manner, making it possible to distinguish and analyze in-
dividual nonlinear products of different orders. The
MTND response simplifies the interpretation of the non-
linear reaction to multitone stimuli by merging the numer-
ous individual distortion spectral components into a single
frequency response of distortion.
Measurement of the frequency-domain Volterra kernels
is also discussed. Plotting these three-dimensional graphs
of distortion of the second and third order is only feasible
if a loudspeaker is characterized by a small level of dis-
tortion (weak nonlinearity). This method quickly loses its
accuracy if the level of distortion is high. High-order Volt-
erra kernels do not have a readily comprehensible graphi-
cal representation.
J. Audio Eng. Soc., Vol. 52, No. 4, April 2004
Of all the methods surveyed and simulated, multitone
testing seems to be the most feasible in the context of
distortion audibility for the assessment of loudspeaker
large-signal performance and nonlinearity measurements.
Nevertheless, harmonic and the traditional two-tone inter-
modulation distortion should not be withdrawn from the
list of standard characteristics. THD is a lower resolution
measure of nonlinearity, but can still be used for the com-
parison of loudspeakers of the same type. Multitone test-
ing is good for both intermodulation distortion measure-
ments and the maximum SPL check. For the latter the
multitone burst should be used. In addition, multitone test-
ing is good for loudspeaker quality control testing.
Setting any boundaries relating objective information
and nonlinear distortion audibility requires extensive com-
puter simulation and involved psychoacoustical tests.
Without such information about the relationship between
objective and subjective parameters, the measurement data
will only be able to tell us that one loudspeaker has more
or less nonlinear distortion. The question of how critical
this difference is from the standpoint of distortion audi-
bility will remain unanswered.
5 REFERENCES
[1] E. Czerwinski, A. Voishvillo, S. Alexandrov, and
A. Terekhov, “Multitone Testing of Sound System Com-
ponents—Some Results and Conclusions, Part 1: History
and Theory,” J. Audio Eng. Soc., vol. 49, pp. 1011–1048
(2001 Nov.); “Multitone Testing of Sound System Com-
ponents—Some Results and Conclusions, Part 2: Model-
ing and Application,” ibid., pp. 1181–1192 (2001 Dec.).
[2] N. Wiener, Nonlinear Problems in Random Theory
(Technology Press, M.I.T., and Wiley, New York, 1958).
[3] O. Dyrlund, “Characterization of Nonlinear Distor-
tion in Hearing Aids Using Coherence Function,” Scand.
Audiol., vol. 18, pp. 143–148 (1989).
[4] J. Kates, “On Using Coherence to Measure Distor-
tion in Hearing Aids,” J. Acoust. Soc. Am., vol. 91, pt.1,
pp. 2236–2244 (1992 Apr.).
[5] Y. Cho, S. Kim, E. Hixson, and E. Powers, “A
Digital Technique to Estimate Second-Order Distortion
Using Higher Order Coherence Spectra,” IEEE Trans. Sig-
nal Process., vol. 40, pp. 1029–1040 (1992 May).
[6] U. Totzek and D. Preis, “How to Measure and In-
terpret Coherence Loss in Magnetic Recording,” J. Audio
Eng. Soc., vol. 35, pp. 869–887 (1987 Nov.).
[7] D. Jensen and G. Sokolich, “Spectral Contamina-
tion Measurements,” presented at the 85th Convention of
the Audio Engineering Society, J. Audio Eng. Soc. (Ab-
stracts), vol. 36, p. 1034 (1988 Dec.), preprint 2725.
[8] R. C. Cabot, “Fast Response and Distortion Test-
ing,” presented at the 90th Convention of the Audio En-
gineering Society, J. Audio Eng. Soc. (Abstracts), vol. 39,
p. 385 (1991 May), preprint 3045.
[9] R. Metzler, “Test and Calibration Application of
Multitone Signals,” in Proc AES 11th Int. Conf. (1992
May), pp. 29–36.
[10] J. Vanderkooy and S. G. Norcross, “Multitone
Testing of Audio Systems,” presented at the 101st Con-
351
VOISHVILLO ET AL.
vention of the Audio Engineering Society, J. Audio Eng.
Soc. (Abstracts), vol. 44, p. 1174 (1996 Dec.), preprint
4378.
[11] P. Schweizer, “Feasibility of Audio Performance
Using Multitones,” in Proc. AES UK Conf. on the Measure
of Audio (1997 Apr.), pp. 34–40.
[12] J. M. Risch, “A New Class of In-Band Multitone
Test Signals,” presented at the 105th Convention of the
Audio Engineering Society, J. Audio Eng. Soc. (Ab-
stracts), vol. 46, p. 1037 (1998 Nov.), preprint 4803.
[13] V. Volterra, Theory of Functionals and of Integral
and Integrodifferential Equations (Dover, New York,
1959).
[14] M. Schetzen, Volterra and Wiener Theories of
Nonlinear Systems (Krieger Publ., Malabar, FL, 1989).
[15] A. M. Kaizer, “Modeling of the Nonlinear Re-
sponse of an Electrodynamic Loudspeaker by a Volterra
Series Expansion,” J. Audio Eng. Soc., vol. 35, pp.
421–433 (1987 June).
[16] M. J. Reed and M. O. Hawksford, “Practical Mod-
eling of Nonlinear Audio Systems Using the Volterra Se-
ries,” presented at the 100th Convention of the Audio
Engineering Society, J. Audio Eng. Soc. (Abstracts), vol.
44, pp. 649–650 (1996 July/Aug.), preprint 4264.
[17] M. J. Reed and M. O. J. Hawksford, “Comparison
of Audio System Nonlinear Performance in Volterra
Space,” presented at the 103rd Convention of the Audio
Engineering Society, J. Audio Eng. Soc. (Abstracts), vol.
45, p. 1026 (1997 Nov.), preprint 4606.
[18] G. Cibelli, E. Ugolotti, and A. Bellini, “Dynamic
Measurements of Low-Frequency Loudspeakers Modeled
by Volterra Series,” presented at the 106th Convention of
the Audio Engineering Society, J. Audio Eng. Soc. (Ab-
stracts), vol. 47, p. 534 (1999 June), preprint 4968.
[19] J. Hilliard, “Distortion Tests by the Intermodula-
tion Method,” Proc. IRE, vol. 29, pp. 614–620 (1941
Dec.).
[20] H. Scott, “The Measurement of Audio Distortion,”
Communications, pp. 25–32, 52–56 (1946 Apr.).
[21] AES2-1984 (r. 2003), “AES Recommended Prac-
tice—Specification of Loudspeaker Components Used in
Professional Audio and Sound Reinforcement,” Audio En-
gineering Society, New York (2003).
[22] IEC 60268-5, “Sound System Equipment—Part 5
Loudspeakers,” International Electrotechnical Commis-
sion, Geneva, Switzerland (2000).
[23] D. Keele, “Method to Measure Intermodulation
Distortion in Loudspeakers,” proposals for the working
group SC-04-03-C, Audio Engineering Society, New York
(2000).
[24] D. B. Keele, “Development of Test Signals for the
EIA-426-B Loudspeaker Power-Rating Compact Disk,”
presented at the 111th Convention of the Audio Engineer-
ing Society, J. Audio Eng. Soc. (Abstracts), vol. 49, p.
1224 (2001 Dec.), convention paper 5451.
[25] E. Zwicker, H. Fastl, and H. Frater, Psychoacous-
tics: Facts and Models, Springer ser. in Information Sci-
ences, 2nd updated ed. (Springer, New York, 1999).
[26] C. T. Tan, B. C. J. Moore, and N. Zacharov, “The
Effect of Nonlinear Distortion on the Perceived Quality of
352
PAPERS
Music and Speech Signals,” J. Audio Eng. Soc., vol. 51,
pp. 1012–1031 (2003 Nov.).
[27] M. Reed and M. Hawksford, “Identification of
Discrete Volterra Series Using Maximum Length Se-
quences,” IEEE Proc. Circuits Dev. Sys., vol. 143, pp.
241–248 (1996 Oct.).
[28] S. Boyd, Y. Tang, and L. Chua, “Measuring Volt-
erra Kernels,” IEEE Trans. Circuits Sys., vol. CAS-30, pp.
571–577 (1983 Aug.).
[29] R. Nowak and B. Van Veen, “Random and Pseu-
dorandom Inputs for Volterra Filter Identification,” IEEE
Trans. Signal Process., vol. 42, pp. 2124–2135 (1994
Aug.).
[30] H. K. Jang and K. J. Kim, “Identification of Loud-
speaker Nonlinearities Using the NARMAX Modeling
Technique,” J. Audio Eng. Soc., vol. 42, pp. 50–59 (1994
Jan./Feb.).
[31] D. Shorter, “The Influence of High Order Products
on Nonlinear Distortion,” Electron. Eng., vol. 22, pp.
152–153 (1950).
[32] E. Geddes, Audio Transducers (Geddlee, 2002),
pp. 236–241.
[33] G. Golub and C. Van Loan, Matrix Computations
(Johns Hopkins University Press, Baltimore, MD, 1996).
[34] A. Voishvillo, “Nonlinear Distortion in Profes-
sional Sound Systems—From Voice Coil to the Listener,”
presented at the Acoustical Conference of the Institute of
Acoustics “Reproduced Sound 17” (Stratford-upon-Avon,
UK, 2001 Nov. 16–18).
[35] J. Bendat and A. Piersol, Random Data: Analysis
and Measurement Procedures (Wiley, New York, 1986).
APPENDIX 1
REFERENCE LOUDSPEAKERS
A1.1 12-in (305-mm) Woofer
The parameters of an experimental 12-in woofer were
obtained from measurements by the Klippel analyzer. In
addition, force factor modulation by the voice-coil current
was modeled using FEM (Fig. 23). The small-signal (rest-
Fig. 23. Bl product affected by flux modulation. Experimental
12-in (305-mm) woofer with overhung voice coil. Coil diameter
75 mm; coil height 38 mm; top plate thickness 15 mm.
J. Audio Eng. Soc., Vol. 52, No. 4, April 2004
PAPERS COMPARISON OF LOUDSPEAKER NONLINEAR DISTORTION MEASUREMENTS

position) parameters of the 12-in woofer used in this work
are given in Table 4. The length of the voice coil is 38 mm,
the diameter is 75 mm, and the thickness of the top plate
is 15 mm. The excursion-dependent parameters Cms, Kms,
Bl, and L1 are shown in Fig. 24. Distortions were simu-
lated for the woofer placed in a sealed 40-liter box.
A1.2 Two 8-in (203-mm) Woofers
The two 8-in woofers used in the experiments and mod-
eling have similar suspension and different motors (Fig.
25). One loudspeaker has a long coil (12 mm) and a short
gap (6 mm), the other has a short coil (6 mm) and a long
gap (12 mm) (Fig. 26). The diameter of both coils is 1.5 in

Table 4

Bl Kms Cms mms Rms Re R2 Le L2 fs Qes Qms Qts Vas

(T ⭈ m) (N/mm) (mm/N) (g) (kg/s) (⍀) (⍀) (mH) (mH) (Hz) (dm3)
20.5 6.55 0.15 232 2.82 5.22 10.4 2.46 1.28 26.8 0.48 13.8 0.46 335

Fig. 24. Excursion-dependent parameters of 12-in (305-mm woofer. (a) Suspension compliance. (b) Suspension stiffness. (c) Bl
product. (d) Voice-coil inductance.

Fig. 25. Reference 8-in (203-mm) woofers used in experiments

and modeling. (a) Loudspeaker A [long coil (12 mm), short gap
(6 mm)]. (b) Loudspeaker B [short coil (6 mm), long gap (12 Fig. 26. Rest positions of voice coils in gap. (a) Loudspeaker A
mm)]. (long coil, short gap). (b) Loudspeaker B (short coil, long gap).

J. Audio Eng. Soc., Vol. 52, No. 4, April 2004 353

VOISHVILLO ET AL. PAPERS

(38 mm). Both loudspeakers did not have dust caps to
prevent any possible artifacts caused by the compression
of the air underneath a dust cap or distortion due to the
turbulent airflow in a pole piece vent. The small-signal
(rest-position) parameters of the loudspeakers are listed in
Table 5.
The nonlinear displacement-dependent parameters for
loudspeaker A are given in Fig. 27, those for loudspeaker
B in Fig. 28. Loudspeaker A (long coil, short gap) has
stronger overall variations of the Bl product and voice-coil
inductance. Using the criterion of the maximum displace-
ment Xmax corresponding to a decrease in the suspension
compliance Cms(x) to 0.12 mm/N, which is to 30% of
its initial value of 0.41 mm/N, the Xmax values of both
loudspeakers were set to 10 mm. At this displacement
the Bl product in loudspeaker A is 2.0 T ⭈ m, which is
22% of its initial value of 9.0 T ⭈ m. The Bl product of
the second driver drops to 3.5 T ⭈ m, which is 47% of
its resting position value of 7.5 T ⭈ m. Such a compar-
atively moderate decrease in the Bl product for loud-
speaker B is explained by the use of an underhung voice
coil.
Using similar suspensions in both loudspeakers and set-
ting identical values of Xmax ⳱ 10 mm helped to compare
the difference in nonlinear distortion in these loudspeakers
caused by the difference in motor parameters.

Table 5

Loudspeaker Bl Kms Cms mms Rms Re R2 Le L2 fs Qes Qms Qts Vas

(T ⭈ m) (T ⭈ m) (N/mm) (mm/N) (g) (kg/s) (⍀) (⍀) (mH) (mH) (Hz) (dm3)
A
Long coil,
short gap 9.0 2.44 0.41 19.2 1.5 5.5 5.4 0.72 0.38 57 0.46 4.6 0.42 33
B
Short coil,
long gap 7.5 2.44 0.41 16.0 1.5 5.3 2.4 0.56 0.29 62 0.60 4.2 0.52 33

Fig. 27. Parameters of loudspeaker A (long coil, short gap) as a function of voice-coil displacement. (a) Suspension compliance. (b)
Suspension stiffness. (c) Bl product; current increments 2 A. (d) Voice-coil inductance.

354 J. Audio Eng. Soc., Vol. 52, No. 4, April 2004

PAPERS COMPARISON OF LOUDSPEAKER NONLINEAR DISTORTION MEASUREMENTS

APPENDIX 2 mechanical losses in the suspension, and Kms(x) is the

suspension stiffness.
The terms
LOUDSPEAKER NONLINEAR MODEL

The nonlinear behavior of the reference loudspeakers dLe共x兲 i2 dL2共x兲 i22

and
was researched numerically using the model described by dx 2 dx 2
a system of two nonlinear differential equations, are the reluctance forces produced by the voice-coil in-
d⌽1共x, t兲 d⌽2共x, t兲 dx ductance Le(x) and the parainductance L2(x). The time de-
U = Rei + + + 关Bl共x兲 + ⌬Bl共x, i兲兴 rivatives of the alternating fluxes ⌽1(x, t) and ⌽2(x, t) are
dt dt dt
expressed as
(24)
d⌽1共x, t兲 di dLe共x兲 dx
dLe共x兲 i2 dL2共x兲 i22 = Le共x兲 + i (26)
关Bl共x兲 + ⌬Bl共x, i兲兴i − − dt dt dx dt
dx 2 dx 2
d2x dx d⌽2共x, t兲 di2 dL2共x兲 dx
= 2 mms + Rms + xKms共x兲 (25) = L2共x兲 + i. (27)
dt dt dt dt dx dt 2
The sound pressure response was calculated by the
where U is the input voltage, i is the voice-coil current, Re
simple far-field half-space expression
is the voice-coil resistance, x is the voice-coil excursion,
⌽1(x, t) is the alternating magnetic flux related to the dy共t兲 Seff␳
voice-coil inductance Le(x), ⌽2(x, t) is the alternating flux p共t兲 = (28)
dt 2␲r
related to the parainductance L2(x), Bl(x) is the force prod-
uct, ⌬Bl(x, i) is the function responsible for the modulation where y(t) is the voice-coil velocity, dy(t)/dt is the voice-
of the flux, induction, and Bl(x) product of the gap, i2 is the coil acceleration, Seff is the diaphragm effective area, ␳ is
current through the parainductance L2(x), mms is the mov- the air density, and r is the distance from the diaphragm to
ing mass of the diaphragm and voice coil, Rms denotes the the observation point.

Fig. 28. Parameters of loudspeaker B (short coil, long gap) as a function of voice-coil displacement. (a) Suspension compliance. (b)
Suspension stiffness. (c) Bl product; current increments 2 A. (d) Voice-coil inductance.

J. Audio Eng. Soc., Vol. 52, No. 4, April 2004 355

VOISHVILLO ET AL. PAPERS

The system of Eqs. (24), (25) was transformed into the is expressed through the loudspeaker parameters as

再 冋 册冎
canonical Cauchy form and solved numerically by the
dZ1共t兲 di2 1 dL2共x兲 dx
classical Runge–Kutta method of the fourth order. The = = R2共x兲i − i2 R2共x兲 −
vector form of the state variables is dt dt L2共x兲 dx dt

Z共t兲 = 兵z1共t兲, z2共t兲, z3共t兲, z4共t兲其T (29)

dZ2共t兲 di
dt
= =
1
dt Le共x兲 冋
U − iRe − i
dLe共x兲 dx
dx dt
− L2共x兲
di2
dt

册
⌽共t, Z兲 = 兵␸1共t, Z兲, ␸2共t, Z兲, ␸3共t, Z兲, ␸4共t, Z兲其 (30)
dL2共x兲 dx
where z1(t) denotes the current i2(t) in the parainductance − i2 − Bl共x兲 − ⌬Bl共x, i兲
dx dt
L2(x), z2(t) stands for the voice-coil current i, z3(t) ac- (35)
counts for the voice-coil displacement x(t), z4(t) is the dZ3共t兲 dx
= =y
voice-coil velocity dx(t)/dt, and ⌽(t, Z), is the vector for- dt dt

再
mulation of the derivatives of the current di2(t)/dt, the
derivative of the voice-coil current di(t)/dt, the voice-coil dZ4共t兲 dy 1 dLe共x兲 i2
= = 关Bl共x兲 + ⌬Bl共x, i兲兴i −
velocity y(t), and the voice-coil acceleration dy(t)/dt. dt dt mms dx 2

冎
The Cauchy vector form of the system of Eqs. (24), (25)
and the initial conditions are dL2共x兲 i22
− − Rmsy − Kms共x兲x .
dx 2
dZ共t兲
= ⌽关t, Z共t兲兴 The function ⌬Bl(x, i) was calculated by the finite-
dt (31)
element method (FEM). The FEM static model of magnet
Z共t0兲 = Z0.
assembly was built and the model of the voice coil was
The integration steps are carried out according to the incorporated. The voice-coil model was ascribed the geo-
following algorithm: metrical dimensions, number of turns, and constant cur-
rent. Using the quasi-dynamic approach, that is, assigning
Zn+1 = Zn + hKn (32) different values of voice-coil current (of positive and
1 negative polarity), the distribution of the gap induction
Kn = 共K共n1兲 + 2K共n2兲 + 2K共n3兲 + K共n4兲兲 (33) was calculated. This procedure was repeated a number of
6
times for different positions of the voice coil. Afterward
and the Bl product was calculated for the corresponding dis-
crete values of the voice-coil current ± Ia, ± Ib ± ⭈ ⭈ ⭈ ± Im
K共n1兲 = ⌽共tn, Zn兲 and the positions of the voice coil ± X1, ± X2 ± ⭈ ⭈ ⭈ ± Xn.

冉 h
2
h
K共n2兲 = ⌽ tn + , Zn + K共n1兲
2 冊 The function ⌬Bl(x, i) approximated the variation of the Bl
product caused by the voice-coil current.

冉 冊
(34) The loudspeaker parameters were measured by the
h h Klippel analyzer and incorporated into the model. Integra-
K共n3兲 = ⌽ tn + , Zn + K共n2兲
2 2 tion of the system of Eqs. (24), (25) was performed using
different input signals to model different measurement
K共n4兲 = ⌽共tn + h, Zn + hK共n3兲兲
conditions. The signal duration and sampling frequency
where h is the time increment. were optimized for a particular signal. The sampling was
The vector linked to the time interval h used in the Runge–Kutta

冋 册
solution of the system (24), (25). The details of the solu-
dZ共t兲 dZ1共t兲 dZ2共t兲 dZ3共t兲 dZ4共t兲 T
tion are not discussed here because they do not have direct
= , , ,
dt dt dt dt dt relation to the subject of this paper.

THE AUTHORS

A. Voishvillo A. Terekhov E. Czerwinski S. Alexandrov

356 J. Audio Eng. Soc., Vol. 52, No. 4, April 2004

PAPERS
Alexander Voishvillo was born and raised in Leningrad
(now Saint Petersburg), Russia. He received a Ph.D. de-
gree in 1987 for work centered on computer optimization
of loudspeaker crossover systems.
From 1977 to 1995 he worked at the Laboratory of
Prospective Research and Development, Popov Institute
for Radio and Acoustics, Saint Petersburg. While at the
Popov Institute he designed loudspeaker systems for
manufacturers and did research work on loudspeakers. He
was responsible for the development of specialized studio
monitors for the Russian Broadcasting Corporation. In
1995 he moved to California, accepting an invitation of
Gene Czerwinski of Cerwin-Vega Inc., to head a new
research and development group. His responsibilities in-
cluded the development of new transducers, as well as
research work on nonlinearity in sound systems and ad-
vanced methods of measurement of nonlinear distortion in
audio equipment. He continued his collaboration with
Gene Czerwinski at Cerwinski Labs, a new R&D com-
pany established in 2002, where he has been working on
the development of original professional high-frequency
transducers and on alternative methods of assessing non-
linearity in audio equipment.
Dr. Voishvillo holds several U.S. patents on new types
of transducers. He is the author of more than 30 publica-
tions on loudspeakers, including the engineering book on
loudspeaker theory and design, High Quality Loudspeaker
Systems and Transducers, published in Russia in 1985 as
well as several publications in the Journal. He is a mem-
ber of the Audio Engineering Society and participates in a
working group on loudspeaker measurements and model-
ing at the AES Standards Committee. He is also a member
of the JAES Review Board.
●
Alexander Terekhov was born in Leningrad (now Saint
Petersburg), Russia, in 1952. He received an M.Sc. degree
in broadcasting and radio communication from the State
University of Telecommunications in 1974.
From 1981 to 1991 he worked as a research associate in
the Laboratory of Prospective Acoustic Research and De-
velopment at the Popov R&D Institute for Radio and
Acoustics, Saint Petersburg. His activities included re-
search on loudspeaker testing and measurements and bin-
aural stereophony. In 1991 he joined Audion Ltd., Saint
Petersburg, as a senior research associate and chief engi-
neer. He came to the United States in 1996 and since 1997
has been employed as an acoustic research engineer at
Cerwin-Vega, Inc., and subsequently at Czerwinski Labo-
ratories. His professional activity is being focused on
acoustic research, the development of new software for
various R&D needs, including loudspeaker measurement
systems, and research on air propagation distortion.
Mr. Terekhov has presented technical papers at Russian
audio conventions and in recent years coauthored several
papers on air propagation distortion published in the Journal.
●
Eugene Czerwinski studied electrical engineering at the
University of Toledo, where he received a B.S.E.E. degree.
He worked as an associate professor of electronics at
the University of Michigan Engineering Research Insti-
tute. Later he was a development engineer at Willys Mo-
tors Electronic Division, where he designed UHF TV
transmitting equipment, studio cameras, and audio sys-
tems. In 1954 he joined the test division of Douglas Air-
craft, where he designed measurement equipment, includ-
J. Audio Eng. Soc., Vol. 52, No. 4, April 2004
COMPARISON OF LOUDSPEAKER NONLINEAR DISTORTION MEASUREMENTS
ing a multichannel sound spectrum analyzer for jet en-
gines. At his next job, with Bendex Pacific, he designed
the first of its kind 10 000-watt germanium solid-state
amplifier for sonar transducers.
His passion for music determined his future life and
career. Coming from a musically deprived childhood, he
was exposed to a live orchestral sound as a young man,
which affected him profoundly. He wanted to be able to
replicate this experience at his time of choosing. While at
Bendex he established Vega Associates, which designed
and manufactured custom high-fidelity systems. In 1973 it
became Cerwin-Vega, Inc. He headed the company for the
next three decades and developed the first quarter-kilowatt
transistor audio amplifier and four-way 18-in loudspeaker
system. In 1964 he started experimenting with live sound
reinforcement for large-venue rock concerts. He worked
with Universal Studios to develop the Sensurround sys-
tem, which received an Academy Award in 1974 for spe-
cial technical achievement in sound. Sensurround was
used nationwide in cinemas to realistically simulate vibra-
tions for the “Earthquake” movie, and was a precursor to
the high-impact theater sound systems later developed by
Lucas Sound and Dolby, Inc. His lifelong interest in music
motivated Mr. Czerwinski to establish in 1989 a nonprofit
recording company, the MAMA (Musical Archives, Mu-
sical Archives) Foundation. The Foundation’s goal is to
preserve the music of culturally significant artists whose
music does not have broad commercial appeal. MAMA
has released over 30 state-of-the-art digital recordings of
jazz and big-band music, and garnered critical acclaim,
including six Grammy nominations and two Grammy
Awards. In 2003 he sold Cerwin-Vega and founded Cer-
winski Laboratories, Inc. At Cerwinski Labs he continues
important research into air-propagation distortion and
multichannel sound reinforcement and reproduction.
Mr. Czerwinski has authored and coauthored numerous
patents on loudspeakers (six patents are currently pending).
●
Sergei Alexandrov was born in Leningrad (now Saint
Petersburg), Russia. He received an M.Sc. degree in elec-
trical engineering from Leningrad University in 1979.
From 1969 to 1978 he worked as a development engi-
neer at the Marine Equipment R&D Institute. From 1978
to 1991 he held a principal position in the R&D group at
Popov R&D Institute for Radio and Acoustics, Saint Pe-
tersburg, where he developed power amplifiers and audio
signal processors for high-fidelity and studio applications.
In 1991 he cofounded and became president and CEO of
Audion Ltd., Saint Petersburg, an audio electronics and
loudspeaker test equipment manufacturing company. In
1996 he came to the United States and held a staff position
as an acoustic R&D engineer at Cerwin-Vega, Inc., until
2002. There he developed original computer-controlled
loudspeaker measurement systems, loudspeaker power
testing devices, and a computer-based system for nonlin-
ear distortion audibility research. He continued his col-
laboration with Gene Czerwinski (the founder of Cerwin-
Vega) at Cerwinski Labs, where he has been working on
loudspeaker magnet system field optimization and non-
standard test equipment design.
Mr. Alexandrov has presented a number of technical
papers at Russian audio and electronics conventions. He
has publications on homomorphic signal analysis, binaural
stereophony, audio signal processing, and loudspeaker
measurements to his credit and coauthored several publi-
cations in the Journal.
357
 ENGINEERING REPORTS

Impedance Compensation Networks for the Lossy

Voice-Coil Inductance of Loudspeaker Drivers*

W. MARSHALL LEACH, JR., AES Fellow

Georgia Institute of Technology, School of Electrical and Computer Engineering, Atlanta, GA 30332-0250, USA

Two simple Zobel impedance compensation networks for the lossy voice-coil inductance
of a loudspeaker driver are described. Design equations for the element values are given, and
a numerical example is presented. The synthesis procedure can be extended to realize general
RC networks which exhibit an impedance that decreases with frequency at a rate of −n
dec/dec, where 0 < n < 1.

At low frequencies the voice-coil impedance is domi-

0 INTRODUCTION
A two-terminal network that is connected in series or in
parallel with a circuit to cause its terminal impedance to be
transformed into a desired impedance is commonly called
a Zobel [1] network. In loudspeaker design a Zobel net-
work consisting of a series resistor and capacitor con-
nected in parallel with the voice coil of a driver has been
described to compensate for the impedance rise at high
frequencies caused by the voice-coil inductance [2]. If the
inductance is lossless, the network can be designed so that
the effective high-frequency impedance is resistive. By
maintaining a resistive load on the crossover network, its
performance is improved. However, the voice-coil induc-
tance of the typical loudspeaker driver is not lossless. In
this case a Zobel network consisting of one resistor and
one capacitor can be used to obtain a resistive input im-
pedance at only one frequency in the high-frequency range
where the voice-coil inductance dominates.
In this engineering report two Zobel networks are de-
scribed, one consisting of two resistors and two capacitors
and the other consisting of three resistors and three ca-
pacitors. Each can be designed to compensate for the lossy
voice-coil inductance of a driver. It is shown that the net-
works can be designed to approximate the desired imped-
ance in an “equal-ripple” sense. Although the approxima-
tion can be improved with the use of more elements, it is
shown by example that the simpler four element network
can give excellent results with a typical driver. The effects
of this network on the responses of second-order and third-
order low-pass crossover networks for a specific driver are
presented.
*Manuscript received 2003 January 24; revised 2003 Novem-
ber 11.
358
nated by its motional impedance. For infinite baffle sys-
tems the low-frequency impedance exhibits a peak at the
fundamental resonance frequency of the driver. In [3] a
modification of the circuit proposed in [2] is described
which provides an additional compensation for this im-
pedance peak. The circuit is also applicable to closed-box
systems. Although the present report concerns impedance
compensation at the high-frequencies where the voice-coil
inductance dominates, the low-frequency compensation
circuit proposed in [3] is reviewed. In addition, a modifi-
cation of this circuit for vented-box systems is given.
It is assumed that the loudspeaker driver is operated in
its small-signal range. Otherwise the voice-coil inductance
becomes a time-varying nonlinear function, its value vary-
ing with diaphragm displacement. This would preclude a
linear circuit analysis and make it impossible to derive the
compensation networks.
The impedance approximation technique presented here
has been used in the design of filters that convert white
noise into pink noise. These circuits exhibit a gain slope of
−3 dB per octave over the audio band. Example circuit
diagrams of such filters can be found in [4], [5], and [6],
but no design equations are given. In [5] the network is
described as one in which “the zeros of one stage partially
cancel the poles of the next stage.” In [7] a similar network
is described to realize an operational-amplifier circuit
which exhibits a gain slope of +4.6 dB per octave over the
audio band. The authors stated that the network compo-
nent values were selected with the aid of a software opti-
mization routine to match the desired slope. An analytical
solution is given here for the design of such networks.
The general impedance compensation theorem de-
scribed by Zobel can be succinctly summarized as follows.
Given an impedance Z1 ⳱ R0 + Z0, let an impedance Z⬘1 ⳱
J. Audio Eng. Soc., Vol. 52, No. 4, April 2004
ENGINEERING REPORTS IMPEDANCE COMPENSATION NETWORKS

R0 + Z⬘0 be connected in parallel with Z1. The condition
that the parallel connection have a resistive impedance
equal to R0 is that Z⬘1 ⳱ R20/Z1. This is a general result that
is not specific to loudspeakers. For completeness, its deri-
vation is given in the following where the notation used is
that for the voice-coil impedance.
circuit has an impedance equal to RE at all frequencies [2].
In this case R1 and C1 form a simple Zobel network, which
cancels the lossless LE from the input impedance of the
driver.
In [10] it is shown that a lossy voice-coil inductance has
an impedance that can often be approximated by

1 IMPEDANCE COMPENSATION CONDITION 冋 冉 冊

ZL共j␻兲 = Le共j␻兲n = Le cos
n␲
2
+ j sin
n␲
2冉 冊册 ␻n (4)

The voice coil of a loudspeaker driver exhibits both a where Le and n are constants. Fig. 2(b) shows the circuit of
series resistance and an inductance. In the following it is Fig. 2(a) with LE replaced with ZL(j␻) and C1 replaced
assumed that the resistance is separated and treated as a with an impedance Z1(j␻). Let Zin be the input impedance
separate element, that is, not a part of the voice-coil in- to the circuit. The source current Is can be written
ductance. Fig. 1 shows the voice-coil equivalent circuit of
a driver in an infinite baffle [8]. The resistor RE and the Vs Vs − V1 Vs − VL
Is = = + . (5)
inductor LE represent the voice-coil resistance and induc- Zin R1 RE
tance. The elements RES, LCES, and CMES model the mo- If Zin ⳱ RE and R1 ⳱ RE, this equation can be solved for
tional impedance generated when the voice coil moves. Vs to obtain
These elements are related to the small-signal parameters
of the driver by the equations [9] Z1 ZL
Vs = V1 + VL = Vs + Vs (6)
RE + Z1 RE + ZL
QMS
RES = R (1) where voltage division has been used to express V1 and VL
QES E
as functions of Vs. This equation can be solved for Z1 to
RE obtain
LCES = (2)

2␲fSQES
QES
CMES =
2␲fSRE
where QMS is the mechanical quality factor, QES is the
electrical quality factor, and fS is the fundamental reso-
nance frequency.
Above the fundamental resonance frequency, the ca-
pacitor CMES becomes a short circuit and the voice-coil
impedance can be approximated by RE in series with LE.
The equivalent high-frequency circuit is shown in Fig.
2(a). A resistor R1 in series with a capacitor C1 is shown
in parallel with the voice-coil impedance. At low frequen-
cies the impedance of the circuit is RE. If the inductor is
lossless, the high-frequency impedance is R1. If R1 ⳱ RE
and R1C1 ⳱ LE/RE, it is straightforward to show that the
Fig. 1. Equivalent circuit of voice-coil impedance.
冋 冉 冊 冉 冊册
R2E R2E n␲ n␲
Z1共j␻兲 = = cos − j sin . (7)
(3) ZL共j␻兲 Le␻ n 2 2
It follows that Zin ⳱ RE if R1 ⳱ RE and Z1(j␻) is given by
Eq. (7). In this case the high-frequency voice-coil imped-
ance is resistive at all frequencies. Note that |Z1(j␻)| ⬀ ␻−n
so that a plot of |Z1(j␻)| versus ␻ on log–log scales is a
straight line with a slope of −n dec/dec. It should also be
noted that Z1(j␻) is the dual of ZL(j␻) scaled by the factor
R2E, which follows from the fundamental principle derived
by Zobel.
2 APPROXIMATING IMPEDANCE
Fig. 3 shows the Bode magnitude plot of an impedance
which exhibits a slope of −n dec/dec between the frequen-
cies f1 and f6. Also shown are the asymptotes of an ap-
proximating impedance which exhibit alternating slopes of
−1 and 0. Four frequencies are labeled between f1 and f6 at
which the slopes of the asymptotes change. In the general
case, let there be N frequencies, where N is even and N ⱖ
4. In this case the number of asymptotes having a slope of
0 is (N − 2)/2. Let k be the ratio of the asymptotic ap-
proximating impedance to the desired impedance at f ⳱ f1.

Fig. 2. (a) High-frequency voice-coil equivalent circuit with two-

element Zobel network. (b) Circuit used to derive Zobel network Fig. 3. Desired impedance and asymptotes of approximating im-
impedance transfer function. pedance versus frequency.

J. Audio Eng. Soc., Vol. 52, No. 4, April 2004 359

LEACH ENGINEERING REPORTS

The desired impedance at f1 is labeled |Z1| in Fig. 3 and is Let Z1( f) be the approximating impedance function. It is
given by given by
|Z1| = Le共2␲f1兲n. (8) k |Z1| 1 + j共 f Ⲑ f2兲
Z1共 f 兲 = × . (15)
The approximating impedance at f1 is labeled k|Z1|. j共 f Ⲑ f1兲 1 + j共 f Ⲑ f3兲
With n, f1, and fN specified, the object is to specify k and 2.2 Case B: N = 6
f2 through fN−1 such that the ratios of each even sub-
Let f1 and f6 be specified. For N ⳱ 6, Eqs. (9)–(11) can
scripted frequency to the odd subscripted frequency to its
be solved to obtain
left are equal and the intersection points (indicated by dots

冉冊
n共1−n兲
on the plot) occur at the geometric mean of the adjacent f 2共2+n兲
frequencies. In this case the lengths of the six dashed k= 6 (16)
f1
vertical lines in Fig. 3 are equal and the asymptotes of the
2 n
approximating impedance approximate the desired imped- f2 = f 2+n f 2+n (17)
1 6
ance in an equal ripple sense between f1 and fN.
1+n 1
It is straightforward to show that the following condi-
f3 = f 2+n f 2+n (18)
tions must hold: 1 6

冉冊 冉冊 冉 冊
1−n 1−n 1−n 1 1+n
f 2 f 2 fN 2 f4 = f 2+n f 2+n (19)
k= 2 = 4 =⭈⭈⭈= (9) 1 6
f1 f3 fN−1
n 2
f5 = f 2+n f 2+n. (20)
f2 = f 1−n
1 f n
3
1 6

The approximating impedance as a function of frequency

f4 = f 1−n
3 f5
n
for this case is given by
⯗ (10)
k |Z1| 1 + j共 f Ⲑ f2兲 1 + j共 f Ⲑ f4兲
fN−2 = f N−3
1−n n Z1共 f 兲 = × × . (21)
f N−1 j共 f Ⲑ f1兲 1 + j共 f Ⲑ f3兲 1 + j共 f Ⲑ f5兲

f3 = f n2 f 1−n
4
2.3 Example Plots
To illustrate the accuracy of the approximating func-
f5 = f n4 f 1−n
6
(11) tions, let the impedance given by Eq. (7) be approximated
⯗ over a three-decade band for the case n ⳱ 0.5. The smaller
fN−1 = f N−2
n
f 1−n the value of n, the poorer the approximation. In the au-
N .
thor’s experience, the value of n for most loudspeaker
Solutions to these equations are given next for the cases drivers is in the range from 0.6 to 0.7. Thus the value n⳱
N ⳱ 4 and N ⳱ 6. 0.5 results in an approximation that is worse than what can
be expected with the typical driver.
2.1 Case A: N = 4 Fig. 4 shows the calculated Bode magnitude plots.
Let f1 and f4 be specified. For N ⳱ 4, Eqs. (9)–(11) can Curve a is the desired impedance. Curve b is the approxi-
be solved to obtain mating impedance for N ⳱ 4. Curve c is the approximat-
ing impedance for N ⳱ 6. It can be seen that the approxi-
冉冊
n共1−n兲
f 2共1+n兲 mating impedance functions ripple about the desired
k= 4 (12)
f1 function over the band of interest with a maximum devia-
1 n tion occurring at the two frequency extremes. Between the
f2 = f 1+n
1
f 1+n
4
(13) two extremes, the maximum deviation is less than it is at
n 1 the extremes because the design equations are derived
f3 = f 1+n
1
f 1+n
4
. (14) from the asymptotes of the approximating function.

Fig. 4. Example plots of desired impedance (curve a) and approximating impedances (curves b, c) versus frequency for n ⳱ 0.5.

360 J. Audio Eng. Soc., Vol. 52, No. 4, April 2004

ENGINEERING REPORTS IMPEDANCE COMPENSATION NETWORKS

3 THE COMPENSATING CIRCUITS The impedance of the circuit is equal to that of Eq. (21) if
3.1 Network A f2 f4
C1 = (33)
Fig. 5(a) shows a circuit consisting of two capacitors 2␲f1 f3 f5 k |Z1|
and one resistor, which can be used to realize the imped-
ance of Eq. (15). The impedance is given by f2 − f3 − f5 + f3 f5 Ⲑ f2
C2 = C1 (34)
f4 − f2
1 1 + s Ⲑ ␻2
Z1共s兲 = × (22)
s共C1 + C2兲 1 + s Ⲑ ␻3 1
R2 = (35)
2␲f2C2
where s ⳱ j␻ ⳱ j2␲f and
1 f3 − f4 + f5 − f3 f5 Ⲑ f4
␻2 = 2␲f2 = (23) C3 = C1 (36)
R2C2 f4 − f2

C1 + C2 1
␻3 = 2␲f3 = . (24) R3 = . (37)
R2C1C2 2␲f4C3

The impedance of the circuit is equal to that of Eq. (15) if

4 COMPENSATING THE LOW-FREQUENCY
f2 DRIVER IMPEDANCE
C1 = (25)
2␲f1 f3 k|Z1|
The impedance Z1(j␻) in Fig. 2(b) is the dual of ZL(j␻)
f3 − f2 scaled by the factor R2E. Following [3], the impedance rise
C2 = C1 (26)
f2 at resonance can be canceled by adding an impedance
Z2(j␻) in parallel with Z1(j␻), which is the dual of the
1
R2 = . (27) motional impedance of the driver scaled by the factor R2E.
2␲f2C2 This impedance consists of a series RLC circuit having the
The circuit of Fig. 5(a) corresponds to Fincham’s more element values [3]
general compensating network in [11].
QES
RS = R (38)
3.2 Network B QMS E
Fig. 5(b) shows a circuit consisting of three capacitors QESRE
and two resistors, which can be used to realize the imped- LS = (39)
2␲ fS
ance of Eq. (21). The impedance is given by
1
1 CS = . (40)
Z1共s兲 = 2␲fSQESRE
s共C1 + C2 + C3兲
共1 + s Ⲑ ␻2兲 共1 + s Ⲑ ␻4兲 The circuit for Z2(j␻) is shown in Fig. 6(a). The preceding
× (28) equations apply to a driver in an infinite baffle. For a
s2 Ⲑ 共␻3␻5兲 + s共1 Ⲑ ␻3 + 1 Ⲑ ␻5兲 + 1
closed-box baffle the element values are given by
where
QEC
1 RS = R (41)
␻2 = 2␲ f2 = (29) QMC E
R2C2
QECRE
1 LS = (42)
␻4 = 2␲ f4 = (30) 2␲ fC
R3C3
1
C1 + C2 + C3 CS = (43)
␻3␻5 = 2␲f3 × 2␲ f5 = (31) 2␲fC QECRE
R2R3C1C2C3
1 1 1 1 where QEC is the closed-box electrical quality factor, QMC
+ = + is the closed-box mechanical quality factor, and fC is the
␻3 ␻5 2␲ f3 2␲ f5
closed-box resonance frequency [12].
R2C2共C1 + C3兲 + R3C3共C1 + C2兲 The circuit for Z2(j␻) for a vented-box baffle is shown
= . (32)
C1 + C2 + C3 in Fig. 6(b). The element values for RS, LS, and CS are

Fig. 5. Circuits for approximating impedance Z1(j␻).

J. Audio Eng. Soc., Vol. 52, No. 4, April 2004 361

LEACH ENGINEERING REPORTS

calculated from Eqs. (38) through (40). It can be shown and their application. The dc voice-coil resistance was
that the elements RP, LP, and CP are given by found to be RE ⳱ 5.1 ⍀. The voice-coil impedance was
␣QES QLRE measured at 62 frequencies between 14.8 Hz and 20 kHz
RP = (44) with an MLSSA analyzer. The data in the range from 1.8
h
to 20 kHz were used to calculate the lossy voice-coil in-
␣QESRE ductance parameters. Calculations on the MLSSA data
LP = (45)
2␲fSh2 yielded the parameters RES ⳱ 26.9 ⍀, LCES ⳱ 38.1 mH,
1 CMES ⳱ 424 ␮F, n ⳱ 0.764, and Le ⳱ 0.0150. Fig. 7
CP = (46) shows the measured magnitude and phase of the imped-
2␲fS␣QESRE ance as circles and the impedance calculated from the
where ␣ ⳱ VAS/VB is the system compliance ratio, QL is equation

冉 冊
the enclosure quality factor at the Helmholtz resonance
1 1 −1
frequency, and h ⳱ fB/fS is the system tuning ratio [13]. ZVC共j␻兲 = RE + Le共j␻兲n + + + j␻CMES
RES j␻LCES
5 NUMERICAL EXAMPLE (47)
One sample of the JBL model 2241H 18-in (0.457-m) shown as a solid line, where ␻ ⳱ 2␲f. The figure shows
professional woofer was selected to illustrate the networks excellent agreement between the measured and calculated

Fig. 6. Compensation circuits for low-frequency impedance rise. (a) Infinite-baffle and closed-box drivers. (b) Vented-box drivers.

Fig. 7. Impedance measured and calculated from Eq. (47) (———) for JBL driver. (a) Magnitude. (b) Phase.

362 J. Audio Eng. Soc., Vol. 52, No. 4, April 2004

ENGINEERING REPORTS IMPEDANCE COMPENSATION NETWORKS

data, thus verifying the calculated values of the driver frequency band from f1 ⳱ 300 Hz to fN ⳱ 20 kHz. Table
parameters. 1 summarizes the intermediate calculations for the two
The element values for the Zobel networks were calcu- networks. Table 2 gives the calculated element values.
lated to compensate for the voice-coil inductance over the Network A is the network of Fig. 5(a). Network B is that
of Fig. 5(b). The element values for the optional network
Table 1. Summary intermediate calculations.
to compensate for the impedance rise at resonance have
Network A Network B the values RS ⳱ 0.858 ⍀, LS ⳱ 11 mH, and CS ⳱ 1460
␮F. It is quite obvious that these values would be imprac-
N 4 6
k 1.24 1.15
tical in a passive crossover network. Indeed, an 11-mH
| Z1| 4.77 ⍀ 4.77 ⍀ air-core inductor would in all probability have a series
f1 300 Hz 300 Hz resistance greater than 0.858 ⍀. For these reasons, the
f2 1.85 kHz 958 Hz impedance Z2(j␻) has been omitted in the following. How-
f3 3.24 kHz 1.37 kHz ever, it would be expected that the element values would
f4 20 kHz 4.38 kHz
f5 6.26 kHz
fall in a more practical range for midrange and tweeter
f6 20 kHz drivers which have a much higher resonance frequency
than the driver considered here.
Table 2. Element values. Fig. 8 shows the magnitude and phase of the voice-coil
impedance with and without Zobel network A. The plots
Network A Network B are calculated from the measured voice-coil data and not
R1 5.1 ⍀ 5.1 ⍀ those predicted by Eq. (47). The plots for network B are
C1 51.2 ␮F 47.4 ␮F not shown because, for all practical purposes, they are not
R2 2.23 ⍀ 5.25 ⍀ distinguishable from those of network A. However, this may
C2 38.5 ␮F 31.6 ␮F not be the case with drivers that have a lower value of n.
R3 2.03 ⍀ To evaluate the effect of the Zobel networks on the
C3 17.9 ␮F
performance of passive crossover networks, the voice-coil

Fig. 8. Impedance of JBL driver with and without Zobel network A. (a) Magnitude. (b) Phase.

J. Audio Eng. Soc., Vol. 52, No. 4, April 2004 363

LEACH ENGINEERING REPORTS

voltage of the JBL driver was calculated for a source volt- Third-order networks are usually designed for a Butter-
age of 1 V rms with second- and third-order low-pass worth response. The crossover frequency is the −3-dB
crossover networks. The crossover frequency was chosen frequency of the network. The element values for the
to be fc ⳱ 800 Hz, which might be a typical value when third-order network in Fig. 9(b) are given by
this driver is used with a midrange horn. The circuit dia-
grams are shown in Fig. 9. Second-order networks are 3RE
L1 = = 1.52 mH (50)
usually designed for critical damping. The crossover fre- 4␲ fc
quency is the −6-dB frequency of the network. The ele-
ment values for the second-order network in Fig. 9(a) are 2
C1 = = 52 ␮F (51)
given by 3␲ fc RE

RE RE
L1 = = 2.03 mH (48) L2 = = 0.507 mH. (52)
␲fc 4␲ fc
1 Figure 10(a) shows the calculated voice-coil voltage for
C1 = = 8.43 ␮F. (49)
4␲ fcRE the second-order crossover network with and without Zo-
bel network A. With the network, the response follows
what would be expected of a second-order crossover net-
work. Without the network, the voltage exhibits a peak at
1.8 kHz that is 16.7 dB greater than the response with the
network. Fig. 10(b) shows the calculated voice-coil volt-
age for the third-order crossover network with and without
Zobel network A. With the network, the response follows
Fig. 9. (a) Second-order crossover network. (b) Third-order what would be expected of a third-order crossover net-
crossover network. work. Without the network, the voltage exhibits a peak at

Fig. 10. Voice-coil voltage of JBL driver with (curve a) and without (curve b) Zobel network A. (a) Second-order crossover network.
(b) Third-order crossover network.

364 J. Audio Eng. Soc., Vol. 52, No. 4, April 2004

ENGINEERING REPORTS
630 Hz that is 10 dB greater than the response with the
network. Above 1.6 kHz the response without the network
lies above the response with the network and exhibits a
slope of approximately −43 dB/dec. The slope with net-
work A approaches −60 dB/dec, which is the correct slope
for a third-order network. Crossover simulations with Zo-
bel network B have been omitted because the results were
almost identical. However, this may not be the case with
drivers having a lower value of n. The plots in Fig. 10
were calculated using the measured voice-coil data and not
that predicted by Eq. (47). The plots show some evidence
of the rise in impedance at the fundamental resonance fre-
quency of the driver. This could be eliminated by the addition
of the circuit in Fig. 6(a) in parallel with the Zobel network.
6 CONCLUSION
The high-frequency rise in the voice-coil impedance of a
loudspeaker driver caused by a lossy voice-coil inductance
can be approximately canceled in the audio band by an RC
Zobel network connected in parallel with the voice coil. The
simplest network consists of two resistors and two capacitors.
More complicated networks have three or more resistors and
three or more capacitors. For a typical driver, the simplest
network can yield excellent results. Because the lossy voice-
coil inductance can cause major perturbations in the per-
formance of crossover networks, the parameters n and Le
should be included in the list of specifications for drivers
as an aid in the design of Zobel compensation networks.
7 REFERENCES
[1] O. J. Zobel, “Theory and Design of Uniform and
Composite Electric Wave Filters,” Bell Sys. Tech. J., vol.
2, pp. 1–46 (1923 Jan.).
[2] R. H. Small, “Constant-Voltage Crossover Network
THE AUTHOR
W. Marshall Leach, Jr. received B.S. and M.S. degrees in
electrical engineering from the University of South Carolina,
Columbia, in 1962 and 1964, and a Ph.D. degree in electrical
engineering from The Georgia Institute of Technology in
1972. In 1964 he worked at the National Aeronautics and
Space Administration in Hampton, VA. From 1965 to 1968
J. Audio Eng. Soc., Vol. 52, No. 4, April 2004
IMPEDANCE COMPENSATION NETWORKS
Design” (Reprint), J. Audio Eng. Soc., vol. 19, pp. 12–19
(1971 Jan.).
[3] A. N. Thiele, “Optimum Passive Loudspeaker Di-
viding Networks,” Proc. IREE (Australia), vol. 36, pp.
220–224 (1975 July).
[4] General Radio Co., Instruction Manual—Type 1382
Random-Noise Generator (1968).
[5] National Semiconductor Corp., Audio Handbook
(1976, 1977, 1980).
[6] P. Horowitz and W. Hill, The Art of Electronics
(Cambridge University Press, Cambridge, MA, 1980).
[7] P. G. L. Mills and M. O. J. Hawksford, “Transcon-
ductance Power Amplifier Systems for Current-Driven
Loudspeakers,” J. Audio Eng. Soc., vol. 37, pp. 809–822
(1989 Oct.).
[8] R. H. Small, “Direct-Radiator Loudspeaker System
Analysis,” J. Audio Eng. Soc., vol. 20, pp. 383–395 (1972
June).
[9] W. M. Leach, Jr., Introduction to Electroacoustics
and Audio Amplifier Design, 3rd ed. (Kendall/Hunt, Du-
buque, IA, 2003).
[10] W. M. Leach, Jr., “Loudspeaker Voice-Coil Induc-
tance Losses: Circuit Models, Parameter Estimation, and
Effect on Frequency Response,” J. Audio Eng. Soc., vol.
50, pp. 442–450 (2002 June).
[11] J. Borwick, Ed., Loudspeaker and Headphone
Handbook, p. 216 (Focal Press–Elsevier, Burlington, MA,
2001).
[12] R. H. Small, “Closed-Box Loudspeaker Systems,
Parts I and II,” J. Audio Eng. Soc., vol. 20, pp. 798–808
(1972 Dec.); vol. 21, pp. 11–18 (1973 Jan./Feb.).
[13] R. H. Small, “Vented-Box Loudspeaker Systems,
Parts I-IV,” J. Audio Eng. Soc., vol. 21, pp. 363–372 (1973
June); pp. 438–444 (1974 July/Aug.); pp. 549–554 (1973
Sept.); pp. 635–639 (1973 Oct.).
he served as an officer in the U.S. Air Force. Since 1972
he has been a faculty member at The Georgia Institute of
Technology, where he is presently professor of electrical
engineering. Dr. Leach teaches courses in applied electro-
magnetics and electronic design. He is a fellow of the Audio
Engineering Society and a senior member of the IEEE.
365
 ENGINEERING REPORTS

Scalable, Content-Based Audio Identification by

Multiple Independent Psychoacoustic Matching*

GEOFF R. SCHMIDT**

Intellivid Corporation, Cambridge, MA 02138-1171, USA

AND

MATTHEW K. BELMONTE**

Departments of Psychiatry and Experimental Psychology, University of Cambridge, Cambridge CB2 2AH, UK

A software system for content-based identification of audio recordings is presented. The

system transforms its input using a perceptual model of the human auditory system, making
its output robust to lossy compression and to other distortions. In order to make use of both
the instantaneous pattern of a recording’s perceptual features and the information contained
in the evolution of these features over time, the system first matches fragments of the input
against a database of fragments of known recordings. In a subsequent step, these matches at
the fragment level are assembled in order to identify a single recording that matches con-
sistently over time. In a small-scale test the system has matched all queries successfully
against a database of 100 000 commercially released recordings.

0 INTRODUCTION
The field of informatics is in the midst of a transforma-
tion from purely textual systems, in which indexing is
driven by tried-and-true methods of string matching, to
multimedia systems, in which measures of similarity are
more dimensionally complex and computationally inten-
sive. As the capacity for information storage has outpaced
developments in algorithms, indexing of pictures and
sounds has been left to rely not on the actual records being
indexed, but rather on file names or other textual labels
attached to these records. Anyone who has made use of
image search engines or peer-to-peer file sharing systems
knows that these labels, or metadata, inevitably fail to
capture essential information. The top matches returned by
a search may turn out to be altered editions related to the
target item (for example, a live concert recording, a remix,
or a cover), or even the wrong item altogether.
To escape this dependence on fallible metadata, search
algorithms are needed that are keyed not on attached text
*Manuscript received 2003 August 12; revised 2003 Decem-
ber 30.
**Formerly with Tuneprint, Inc., Cambridge, MA 02139-
4056, USA.
366
but on the actual data being indexed. The complexity in-
herent in such algorithms is a product of the fact that
perceptual similarities appear not so much in a media file’s
raw data as in its many derived properties. Similarities
apparent to the human senses are seldom evident in com-
parisons of the data’s raw bytes. Images whose actual
pixel values are utterly different from each other may
nonetheless look alike to the human visual system, and
sounds whose time-series data are uncorrelated may none-
theless sound alike to the human auditory system. Samples
may vary along multiple perceptual axes, making the
search space high-dimensional and therefore making near-
est-neighbor searches exponentially more difficult. A
simple example of this dimensionality problem is a time–
frequency representation of a sound, in which the number
of dimensions is the product of the number of time steps
and the number of frequency bands. In order for the prob-
lem of comparing media files to be rendered tractable, the
dimensionality of the inputs must be reduced. The process
of deterministically computing a relatively low-dimensional,
unique identifier for a media file is known as the fingerprint-
ing problem. Fingerprinting in the audio domain, in particu-
lar, has received a great deal of attention due to its immediate
applications in archive indexing and searching, automated
broadcast monitoring, and digital rights management.
J. Audio Eng. Soc., Vol. 52, No. 4, April 2004
ENGINEERING REPORTS AUDIO IDENTIFICATION BY MULTIPLE INDEPENDENT PSYCHOACOUSTIC MATCHING
The audio fingerprinting schemes published to date ex-
tract information by applying frequency-domain analyses
within narrow temporal windows. Some of these methods
use a straightforward time–frequency representation as in-
put to further processing steps [1]–[6] whereas others use
the Fourier spectrum to compute derived measures such as
modulation frequency [7], measures of spectral shape and
tonality [8], [9], MPEG-7 feature sets [10], or hash bits
based on frequency-specific temporal changes [11], [12].
One published method extracts individual notes and then
applies conventional string-matching algorithms to se-
quences of these notes [13], though of course such a
method does not allow for complexities such as superpo-
sitions of various instruments or vocals. All of these sys-
tems either depend on supervised learning algorithms ap-
plied to fairly raw time–frequency representations, or
decide a priori what specific spectral features or measures
will be relevant. In the former case, perceptually relevant
information is submerged in a large corpus of data, making
statistical learning algorithms liable to discriminate on the
basis of perceptually irrelevant features. In the latter case,
feature extraction discards a great deal of useful informa-
tion along with the irrelevant details. Either way, classifi-
cation is degraded.
In addition to this loss of perceptual information within
time points, many existing techniques do not make full use
of information on changes in a recording’s perceptual
qualities across time points. A single recording may
evolve temporally through many different styles, tempos,
and timbres. Previous audio fingerprinting methods have
suffered from a needlessly exclusive view of time and
frequency representations, collapsing localized frequency-
domain features across long intervals of time. This strat-
egy accomplishes a great deal of data reduction, at the
expense of blurring out short-lived properties that could be
useful for identification. The Muscle Fish system [8], for
example, computes feature vectors for a large set of
closely spaced time frames within a recording, but then
retains only the mean, variance, and autocorrelation of
these feature vectors over the entire recording. This strat-
egy works well for brief samples such as sound effects [8]
but is unlikely to scale well to temporally extended re-
cordings. A system described recently by Sukittanon and
Atlas [7] adopts a similar approach, computing subband
modulation frequencies in each frame and then preserving
only the centroids of these features across frames. A sys-
tem described by Burges et al. [6] operates only on brief
clips, basing its classifications solely on the one frame that
differs least from a frame in the database, without using
information from surrounding frames. Other systems [2],
[4] pool feature vectors across the entire recording into a
histogram of vector-quantization bins, destroying temporal
sequence information as the data from the individual vec-
tors are summed into the histogram. Another system based
on vector quantization [9] sums the error between feature
vectors and vector quantization codebook entries, and then
classifies the recording as a whole on the basis of the
codebook entry whose summed error is least. Here again,
information on the recording’s time course is lost in the
summation.
J. Audio Eng. Soc., Vol. 52, No. 4, April 2004
One way around the problem of temporal blurring in
pooled data is to sum not the match scores themselves but
rather the outputs of some nonlinear “squashing function”
applied to the match scores. Good matches of short-lived
features will thus be given a disproportionate effect on the
match quality of the recording as a whole. A system de-
signed by Papaodysseus et al. [5] adopts this approach
using a simple step function as a squashing function, that
is, matches at each time point are thresholded, and the
summed match score is incremented only if the match at
the current time point is above threshold. With this ap-
proach also, though, useful data are lost, since information
on partial, subthreshold matches within individual time
points has no effect on matching the recording as a whole,
even when such partial matches occur at a large number of
time points.
A difficulty in applying information from partial
matches is the time complexity of searching for these
matches. Exact matches, on the other hand, are much more
efficiently identified. Quantization of the signal’s features
will produce some number of exact matches with identi-
cally quantized records in the database. The database re-
cordings that generated these exact matches can then be
searched for approximate matches. Constraining the
search space in this way renders the approximate search
problem tractable. This strategy has been applied in a
hashing system based on temporal differences in subband
amplitude differences (that is, a double differentiation in
frequency and time) [11], [12]. Such a search method pre-
serves information across time points, though it remains an
open question whether this method of time–frequency dif-
ferentiation can be improved on for extracting perceptu-
ally relevant information within time points.
Losses of useful data within and across time points are
likely a major reason why fingerprinting schemes in gen-
eral have suffered from error rates that would be unac-
ceptable in any large-scale system. Though more advanced
systems (such as [12]) promise improved results, most of
the methods cited feature rates of successful matching in
the range of 90 to 99%. When matching against a database
of hundreds of thousands or even millions of recordings,
even 99% is unacceptable. Worsening the outlook for scal-
ability is the fact that many of these error rates arise from
tests against small databases on the order of hundreds [8],
[1]–[4] or thousands [13], [9], [7] of recordings, or within
restricted musical genres [1], [3]. In order to improve per-
formance, strategies must be developed to extract selec-
tively the perceptually relevant information within time
points, and to preserve this information across time points.
For both of these goals one can look to human neurobiol-
ogy as a model.
The goal of preservation of feature-specific information
across time points has been addressed in a biologically
motivated neural network model proposed by Hopfield
and Brody [14]. This model recognizes audio signals via a
massively parallel network of independent feature detec-
tors whose outputs decay with various time constants. The
recognition signal consists of a selectively weighted sum
of these many time-dependent, feature-dependent mea-
sures. Such a network of multiple independent feature de-
367
SCHMIDT AND BELMONTE
tectors is the parallel-processing equivalent of a serial
strategy involving multiple independent matching of fea-
tures at a series of time points, followed by an evaluation
of the match results for temporal consistency. This strat-
egy makes the recognition problem tractable by separating
the problem of matching within time points from the prob-
lem of matching across time points. Matching within time
points produces in general some scalar measure of good-
ness of fit. Matching across time points then amounts to an
instance of the well-known problem of weighted linear
regression, where the time indices of an unknown input
recording are regressed against the time indices of a ref-
erence recording using these within-time-points measures
as weights, and a regression line of unit (or near-unit)
slope indicates a match. This method of weighted linear
regression has been applied to audio data by Schmidt [15],
[16] and independently by Wang [17].
The choice of a method for extracting perceptual infor-
mation within time points depends on the context for
which the audio fingerprinting system has been designed.
While some systems [4] only classify the input as to mu-
sical genre or class, others rank better and worse matches
within a class [8], or identify single best matches [13], [2],
[9], [5], [7]. Among the systems designed for individual
identification, variously inclusive or expansive criteria can
be established. The most narrow and least useful sense of
identity is the simple equality of two signals. In this case,
byte-for-byte comparison suffices and no fingerprinting is
needed. At the opposite extreme, some applications may
need to establish broad identity, retrieving sounds that
contain similar sequences or thematic elements but differ-
ent instrumentals or voices (such as remixes or covers).
Perhaps most useful for applications of digital rights man-
agement, though, is an audio indexing system that identi-
fies an input as one record within a database of releases,
after that input has perhaps been distorted by slight varia-
tions in playback speed, by lossy compression, or by trans-
mission through a bandwidth-limited system. Such an in-
dexing system would identify recordings that sound the
same to a human listener, and would differentiate record-
ings that contain obvious differences.
This criterion of equivalence to a human listener leads
to a natural approach to enhancing perceptually relevant
information within time points: if the software is made to
model human auditory processing, then features not im-
portant in human auditory discrimination will be lost, and
features that contribute to discrimination will be retained.
As a result, the criterion of equivalence to a human listener
can be attained without any assumptions as to what par-
ticular features are relevant. A classifier based on human
auditory modeling is in a way an elaboration of time–
frequency methods that partition the audible spectrum into
frequency bands akin to the critical bands of the human
cochlea [2], [4], [5]. One such method has achieved 99.8%
recognition by applying dimensionality reduction and su-
pervised learning after accounting for frequency-specific
human auditory perceptual thresholds [6], though more
complex psychoacoustic phenomena such as spreading
and compression are not accounted for.
368
ENGINEERING REPORTS
In order to preserve information across time points, we
adopt the philosophy that many approximate tests of
matching are better than a single, make-or-break test.
Other systems have constructed databases whose elements
are whole recordings, and this strategy leads inevitably to
the problem of pooling data across time points. In contrast,
the system discussed in this study, known as Tuneprint,
matches each short fragment of a recording against a large
database consisting of every individual fragment of every
recording loaded. Fast vector quantization within such a
large database is inevitably fallible, and thus many frag-
ments within an input recording will produce incorrect
matches. A significant portion of matches, however, will
be correct, and it is this redundancy across time points that
is the key to the Tuneprint algorithm: if significantly many
fragments from the input produce matches that turn out to
be the temporally corresponding fragments from a single
recording, the input can be identified as this recording.
Using a sophisticated psychoacoustic model as the front
end to this fragment-based analysis, Tuneprint has
achieved a high success rate on a database of 100 000
commercial music releases.
1 THE ALGORITHM
1.1 General Considerations
The identification algorithm can be conceptualized as
the combination of a perceptual model, which eliminates
features that are not significant to a human listener, a
fragment function, which matches each of the input’s in-
dividual temporal fragments against fragments in the da-
tabase, and an assembly function, which puts together re-
sults from the fragment function over time and evaluates
their consistency. In general, a fragment function takes an
input recording and a temporal offset within that record-
ing, and produces a set of triples, each consisting of a
recording in the database, a temporal offset within that
recording, and a distance measure or some other quantifi-
cation of match quality. An assembly function takes a
series of the fragment function’s outputs over time, and
produces a set of outputs each of which associates a re-
cording in the database with a confidence level with which
the input matches that recording.
The goal in defining the fragment function is not so
much to maximize accuracy as to maximize the informa-
tion garnered per unit of computational resources spent.
Depending on how the fragment function is implemented,
the limiting resource may be input–output bandwidth,
memory bandwidth, network bandwidth, or CPU time. It is
expected that a fragment function will be noisy, even re-
turning no information at all for some particularly difficult
cases. The assembly function is able to extract a signal
from this noisy output by exploiting the constraint that any
genuine match must be consistent over an interval of time.
Since matching is evaluated independently for each
fragment, it is possible for a match to occur between any
subinterval of a recording submitted for identification and
any matching subinterval of a recording in the database.
This ability to construct independent matches on subinter-
vals makes the Tuneprint system robust to truncation of
J. Audio Eng. Soc., Vol. 52, No. 4, April 2004
ENGINEERING REPORTS AUDIO IDENTIFICATION BY MULTIPLE INDEPENDENT PSYCHOACOUSTIC MATCHING
audio, to the addition of silent intervals, and to momentary
glitches such as sometimes occur during radio or stream-
ing transmission. Useful results can be obtained even from
less than a second of input.
1.2 Psychoacoustic Modeling
As a front end to Tuneprint’s fragment function, audio
recordings are transformed by a psychoacoustically based
model of human hearing. Input is sampled at 44.1 kHz,
either from raw CD content or by playback from a com-
pressed format. In the case of a stereo recording, left and
right channels are mixed to mono. Intervals of 185.715 ms
(8190 samples) are normalized for playback volume by
subtracting the mean and then scaling to the interval’s
maximum excursion or to one-sixteenth of the playback
medium’s dynamic range, whichever is greater. (This dy-
namic range criterion prevents silent intervals from be-
coming high-amplitude noise.) The interval is multiplied
by a Hann window, padded with a single zero on each end,
and then Fourier-transformed. A power spectrum with a
resolution of 5.38 Hz is extracted from the Fourier trans-
form, over the frequency range from 253 to 12 500 Hz.
Frequencies outside this range are discarded so that
matches cannot depend on them. We have found this strat-
egy effective at improving the identification of band-
limited transmissions while having no negative effect on
the identification of full-bandwidth recordings. The power
spectrum is transformed from a hertz scale to a Bark scale
by linear interpolation using the Bark frequency values
given in [18]. This transformation yields a power spectrum
extending from 2.53 Bark to 23.17 Bark in 128 discrete
steps, energy from multiple frequency bins at the high end
of the spectrum being summed into single Bark bins.
Transformation to the Bark scale sets the stage for the
computation of frequency spreading. In the human ear,
mechanical properties of the basilar membrane and coarse
coding within the cochlear nucleus cause a single-
frequency input to excite neurons encoding a range of
frequencies, with a central peak of excitation occurring at
the input frequency. This spreading of neural excitation
across frequencies underlies the psychoacoustic phenom-
enon of masking, in which a low-amplitude tone, at a
frequency near that of a higher amplitude tone or a band of
noise, cannot be resolved [18]. Most applications based on
human perceptual modeling (for example, MPEG layer 3
[19]) compute an intensity threshold below which a
masked sound will not be heard. Since thresholds differ
depending on whether the masker is a pure tone or noise,
this thresholding method has the disadvantage of relying
on rather arbitrary measures of spectral flatness as indices
of tonality.
An alternative to modeling the masking threshold is to
model the frequency spreading itself. With this approach,
the masked threshold is not explicitly computed. Instead, it
arises from decreases in discriminability due to spreading
of the spectrum. The computational distinction between
noise masking and tone masking also is obviated [20]: the
same model is used in both cases, and the difference arises
in the model’s behavior in response to masking inputs
whose levels are fairly uniform across frequencies (noise
J. Audio Eng. Soc., Vol. 52, No. 4, April 2004
maskers) versus those whose levels are frequency-specific
(tone maskers). As specified in [20], we spread each Bark
band with a left-sided rolloff of 31 dB/Bark and a right-
sided rolloff of 22 dB/Bark + (230 Hz/Bark)/␯, where ␯ is
the frequency (in hertz) of the masker. We apply an in-
tensity compression factor ␣ of 0.8, making the com-
pressed sum of excitations greater than the linear sum.
Algorithmic details on the application of this compression
factor and the computation of spreading are presented in
[20]. In addition to accounting for masking, this spreading
of spectral peaks makes the identification robust to dis-
cretization errors that may arise from small variations in
playback speed.
Following spreading and conversion to an intensity (dB)
scale, the minimum audible field (MAF) [21], expressed in
dB as a function of Bark frequency, is subtracted from the
signal. Bark frequency bands in which this subtraction
yields a negative result are zeroed, whereas nonnegative
results are transformed according to a perceptual loudness
measure based on that defined in [22], depending on Bark
frequency z and frequency-specific intensity Iz:
0.7 + 0.4 Ⲑ Bark ⭈ 共2.5 Bark − max关2.5 Bark, min共3 Bark, z兲兴兲
冋
⭈ 共Iz − 100 dB兲 + 100 dB + 8.5 dB ⭈ 1 −
1+e
1
册
−0.09 Ⲑ dB⭈共Iz−60 dB兲
.
This step completes the psychoacoustic transformation, an
example of which is shown in Fig. 1. The resulting output
changes fairly slowly across samples (Fig. 2), making
Tuneprint robust to temporal frame shifts.
As a postprocessing step, the spectrum is high-pass
filtered by subtracting from each point the linear trend in
the 6-Bark interval centered on that point. (In the case of
points that lie within 3 Bark of the spectrum’s upper or
lower edge, the detrending window is narrowed accord-
ingly.) Although this last filtering step may seem to di-
verge from the goal of modeling human perception, we
find it useful in practice since it removes band-limited
intensity offsets that arise from equalization (see example
in Fig. 3). Such equalization can be produced deliberately,
but is more often an unintentional consequence of the
limited frequency response of amplifiers, loudspeakers,
microphones, or transmission systems.
The end result of all these transformations is a feature
vector whose 128 components represent high-pass-filtered
perceptual intensity as a function of Bark frequency, in a
brief interval around the time point of interest. In order to
pair this frequency-domain information with some local
temporal information, the psychoacoustic transformation
is repeated in adjoining fragments, one immediately pre-
ceding and the other immediately following the fragment
of interest. These temporal offsets produce a total of three
128-dimensional feature vectors, which are concatenated
to form a 384-dimensional feature vector. The time point
of interest is advanced through the recording in half-
fragment steps of 92.8575 ms, so each time point is in-
corporated into two different 128-vectors, and each 128-
vector is used in three different 384-vectors.
Especially after frequency spreading and duplication
across time, these feature vectors contain a great deal of
369
SCHMIDT AND BELMONTE ENGINEERING REPORTS

redundant information. In order to avoid the problem of

searching in a very high-dimensional space, the 384-
vectors are reduced to 8-vectors by principal components
analysis (PCA), a method also applied in a system by
Burges et al. [6]. PCA [23] is a standard method for re-
ducing dimensionality by exploiting correlations between
dimensions. In 3-space, for example, a set of vectors
whose coordinates were perfectly uncorrelated might form
a spherical volume of scattering, while a set of vectors
whose coordinates were perfectly correlated would form a
line. A correlation much greater than 0 but somewhat less
than 1 would describe a cylindrical volume, the long axis
of which would capture most of the variance in the data. If
one wanted to reduce the dimensionality of the data from
3 to 1, PCA could be applied in order to discover this axis
of greatest variation, and all of the original 3-vectors could
then be projected onto it. Fig. 2. Evolution of psychoacoustic transformation across suc-
More formally, PCA operates on an N-dimensional cessive fragments. Bark frequency is on horizontal axis, loudness
space by computing from a set of training data the N × N on vertical axis, and time on depth axis. In these half-
covariance matrix where element (i, j) is the covariance of overlapping, 93-ms fragments, psychoacoustic features arise and
the ith dimension with the jth. The eigenvectors of this disappear gradually over time.
matrix are the principal components, and for each of these
principal components the corresponding eigenvalue de-
scribes the amount of variance captured. Reduction from
384 dimensions to a basis of eight dimensions thus can be
accomplished by computing the eigenvectors of the 384 ×
384 covariance matrix and extracting the eight eigenvec-
tors whose eigenvalues are largest. The transformation
matrix from the original 384-space to a variance-
maximizing 8-space then consists of these eight vectors, or
principal components.
The training data for our PCA consist of all of the
384-dimensional feature vectors computed from nonover-
lapping groups of three fragments, that is, from samples at
intervals of 557.145 ms. The eight basis vectors computed
in this analysis (Fig. 4) account for 46.5% of the variance Fig. 3. Comparison of two inputs, identical except for equaliza-
in our database of 100 000 recordings. Two of these basis tion profiles. Note almost exact similarity in local psychoacoustic
vectors capture temporal information, changing sign features and frequency-dependent offset in absolute amplitudes.

Fig. 1. Steps in psychoacoustic transformation. Power is transformed to a Bark scale and spreading is applied to mimic the mechanical
properties of the human cochlea, implementing the psychoacoustic phenomenon of masking. Amplitude of the spread spectrum is
converted to dB, auditory threshold is subtracted, and resulting levels are converted to an arbitrarily scaled, dB-like measure of
perceptual loudness.

370 J. Audio Eng. Soc., Vol. 52, No. 4, April 2004

ENGINEERING REPORTS AUDIO IDENTIFICATION BY MULTIPLE INDEPENDENT PSYCHOACOUSTIC MATCHING

across each group of three successive fragments, and the within a given partition times the information contained in
other six capture frequency-domain information that is such a partitioning:
fairly constant across successive fragments. Although pro-

兺 冋|database| ⭈ − log 冉 | database| 冊册.

jecting onto this compact basis preserves slightly less than |partition| | partition|
half of the variance in the total corpus of recordings, our 2
partition
framework of multiple independent comparisons makes
the system robust to the resulting increase in match fail-
ures. An eight-component basis greatly reduces storage High entropy decreases processing time. One factor in this
requirements in comparison to larger bases, and even dou- decrease is that more partitions allow a database to be
bling the dimension of the basis would account for only an spread over many separate back-end servers. More signifi-
additional 10.4% of the sample variance (Fig. 5). cantly, though, even when many partitions are allocated to
a single server, high entropy means that each partition
contains fewer search candidates.
1.3 Fragment Analysis
Vectors are stored in the database as eight 4-byte IEEE
floating-point numbers, with 20 additional bytes associat-
ing the vector with its source recording and a temporal
offset within that recording. With an average recording
length of about 4 min, the 100 000 recordings in the da-
tabase occupy 12.2 Gbyte of storage. In order to attain
high throughput for match queries, this storage is split
across seven separate back-end servers, each holding a
1.75-Gbyte slice of the master database. Vectors are quan-
tized into partitions using the generalized Lloyd algorithm
[24], [25], and only the partition whose centroid has the
closest Euclidean distance to the query vector is searched.
One or more front-end index servers accept queries and
dispatch each of them asynchronously to the back-end
machine that holds the appropriate partition (Fig. 6). In
general, the number of back-end machines required for a
particular application is a function of both the size of the
database (given a limit on memory per single PC bus) and
the volume of queries per unit time. These two factors
scale somewhat independently: a small, heavily loaded
Tuneprint system can split a small amount of memory
across a large number of CPUs, whereas a large Tuneprint
system with a smaller query volume can concentrate more
memory in a smaller number of machines.
The partitioned database can be described by two in-
versely related measures: entropy and partition error. The
entropy of the system is the expected amount of informa- Fig. 5. Variance accounted for as a function of number of prin-
tion necessary for its description—in this case the sum, cipal components preserved. Eight components account for
over all partitions, of the probability that a vector will fall 46.5% of variance. From there, curve increases more gradually.

Fig. 4. Basis vectors computed by principal-components analysis. Each input is a concatenation of three 128-dimensional feature
vectors in successive fragments; frequency-selective components therefore replicate the same features three times over. Note broad
frequency selectivity of components 1 through 3, and time dependence of components 4 (decay and attack) and 5 (beat). Higher order
components show a spiky pattern of frequency selectivity, perhaps incorporating information from notes and harmonics. Since basis
vectors are normalized to unit length, magnitudes are arbitrary and are not specified here.

J. Audio Eng. Soc., Vol. 52, No. 4, April 2004 371

SCHMIDT AND BELMONTE
The tradeoff is that with high entropy comes a high
proportion of incorrectly partitioned queries. In the limit-
ing case, in which a query vector is identical to one of the
vectors in the database, of course, the query will neces-
sarily be allocated to the same partition as its best match.
However, if the query vector lies some distance from its
best match, there is some risk that the query and the best
match will fall on different sides of a partition boundary.
In such a case the best match will never be identified,
because the partition that contains it will not be searched.
The optimal partitioning scheme maximizes entropy with
the constraint that the rate of query partitioning errors is
low enough to produce good matches at a preponderance
of time points. The balance between these two factors of
high entropy and low partitioning error depends on the
average distance between queries and their best matches:
again, in the limiting case where this distance is 0, each
vector in the database could be its own partition (and the
entire problem could be handled using string-matching
algorithms).
Within the selected partition we consider only the
neighborhood N of vectors whose Euclidean distance from
the query vector q does not exceed a threshold distance
dlimit. The theme behind this computation is to do only as
much work as is necessary to produce the desired number
of best matches. In general, the algorithm operates by
considering successively larger subspaces of the query
space. We begin by considering the one-dimensional sub-
space defined by the first principal component. (Recall
that the principal components are the basis in terms of
which the query space is represented; projections onto the
subspaces defined by these components therefore have
zero computational cost, being implemented simply by
discarding the higher order dimensions.) To facilitate dis-
tance computations within this one-dimensional subspace,
the contents of the database partition are maintained in
sorted order according to the value of the first principal
component. The set N0 of all vectors that lie within a
distance ±dlimit of the query vector along this principal
component can thus be computed with an efficient search.
Fig. 6. Block diagram of Tuneprint processing. Queries are dis-
patched to appropriate back-end server.
372
ENGINEERING REPORTS
Since distance along the principal component is a lower
bound for actual distance, N ⊆ N0.
In order to exclude those vectors that are in N0 but not
in N, we organize the elements of N0 into a heap keyed on
lower bound distance from the query vector. A heap, as
used in this case, is a binary tree in which the value of
every child node is greater than or equal to the value of
that child’s parent node. The root node in such a tree is
termed the “top” of the heap, and it necessarily contains
the heap’s minimal value. We consider this minimal value,
summing into it the squared distances along the next ⌬n
principal components. The addition of these extra dimen-
sions tightens the lower bound on the vector’s actual dis-
tance from the query. In the case in which this addition
causes the value at the top of the heap to grow larger than
the distances lower down in the heap, we swap values
down the tree in order to reestablish the heap property.
We continue this process, on each iteration increment-
ing the number of dimensions in the top vector’s distance
estimate by ⌬n, and then reestablishing the heap property
on the basis of this increased distance estimate. When a
vector appears at the top of the heap for which all the
dimensions in the query space have been considered, that
is, for which the lower bound distance is identical to the
actual distance, we identify that vector as the next match,
and delete it from the heap. This incremental computation
of distances allows us to perform only the computation
that is necessary in order to identify the closest matches.
In the case of the eight-dimensional space around which
the current implementation is built, ⌬n ⳱ 7, and thus all
the higher dimensional distance computation is done at
once. However, this incremental algorithm has improved
performance in tests with higher dimensionality, and may
prove useful if scaling of the system demands an increase
in the number of basis vectors. Although the search algo-
rithm without incremental distance computation is O[D
|N0| log(|N0|)] for N0 vectors in a space of D dimensions, in
practice constant factors are such that this time bound is an
improvement on O(D|N0|) exhaustive search. The algo-
rithm is specified formally in Fig. 7.
Let D be the dimensionality of the query space;
Let q ⳱ (q0 . . . qD−1) be the query vector within this space;
Let k be the number of matches desired;
for {v | (␯0 − q0)2 ⱕ d2limit}
dim(v):⳱ 1;
d2(v):⳱ (␯0 − q0)2;
initialize N as a heap keyed on d2;
MATCH :⳱
;
do | MATCH | ⫽ k ∧ | N | ⫽ 0 →
Let v be the vector at the top of heap N;
if dim(v) ⳱ D →
MATCH :⳱ MATCH ∪ {v};
delete v from N
□ dim(v) < D →
d2(v) :⳱ d2(v) + ∑i⳱dim(v)
dim(v)+⌬n−1
(␯i − qi)2;
dim(v) :⳱ dim(v) + ⌬n;
reestablish N as a heap on d2
fi
od
Fig. 7. Algorithm for incremental search of query neighborhood.
J. Audio Eng. Soc., Vol. 52, No. 4, April 2004
ENGINEERING REPORTS AUDIO IDENTIFICATION BY MULTIPLE INDEPENDENT PSYCHOACOUSTIC MATCHING
1.4 Assembly
Using this procedure we obtain the closest 10 matches
for each time step of the query recording. Each of these
matching database entries is tagged with its source record-
ing and its temporal offset within that source recording.
Assembling the information from these multiple indepen-
dent matches is a matter of detecting whether a prepon-
derance of them originates from a single recording, with
temporal offsets corresponding to their sequence within
the query recording.
For each time step in the query, we first scale the results
for each recording in the database that has produced more
than a single match for this time step. (Singleton match
sets are likely spurious—the results of partition errors and
of collisions in the database—and are discarded.) If M is
the number of matches originating from a particular
matching recording, di is the distance between the query
vector and the match vector for the ith match, and (dmin,
dmax) is the interval covering all the di’s, then the score for
each match is defined as
di + dmin − 2dmax
S= .
共兺 M−1
i=0 兲
di + Mdmin − 2Mdmax
This expression scales the best match to 1 and the worst to
1/2, with the intermediate scores being a linear function of
the match distance. The resulting scores are pooled across
time steps for each matching recording. These match
scores over time can be visualized as a two-dimensional
plot, with query time on one axis, match time on the other,
and the intensity of each pixel corresponding to the score
of the match against the query at the corresponding time
point. A strong match thus appears as a bright diagonal
line (Fig. 8) [16].
The top candidates in terms of number of time steps
matched are considered further. Linear regression on the
temporal offsets from each match, weighted by the match
Fig. 8. Graphical representation of a successful match. Horizon-
tal axis is temporal offset within a query recording; vertical axis
is temporal offset within a candidate match in database. Intensity
of each pixel (i, j) represents distance between vector represent-
ing ith sample of query recording and vector representing jth
sample of database recording. Thus a perfect match appears as a
bright diagonal line, matches against refrains appear as diagonal
line segments offset from main line, and absence of a match
appears as a dark field.
J. Audio Eng. Soc., Vol. 52, No. 4, April 2004
scores, establishes the slope of the regression line between
query time and match time—in other words, the playback
speed. In most cases this slope will be 1. However, it is
possible for the playback speed to differ slightly from 1,
for example, in the case of a broadcaster trying to pack as
many tracks as possible into an hour of air time. In appli-
cations in which the playback speed is known to be 1, this
regression step can be skipped.
Linear regression could also be applied to estimate the
intercept, that is, the length of the recording’s initial si-
lence or the amount by which the recording has been trun-
cated at its beginning. Such an estimate, however, would
be perturbed by the repetition of themes and refrains,
which produce short sequences of matches with very dif-
ferent temporal offsets (see Fig. 8). In order to avoid such
perturbation, we estimate the intercept by sliding the re-
gression line through the match plot and computing the
total number of matches within a five-point window cen-
tered on the regression line. The intercept that maximizes
this local match total is taken as the actual temporal offset,
and the maximal total itself is the match score of the
recording as a whole. All the recordings under consider-
ation are then ranked in order of this match score.
2 PERFORMANCE
For a preliminary evaluation of Tuneprint’s perfor-
mance on real-world data, a test corpus was created of
randomly selected recordings that were publicly available
as MPEG-1 layer 3 files on a peer-to-peer network. In
cases in which listening tests indicated that one of these
downloaded recordings was not in our database, or was a
duplicate of a recording already in the test set, the record-
ing in question was deleted from the test set. These dele-
tions left a test set of 73 unique recordings. In all 73 cases,
Tuneprint identified the test recording correctly as the da-
tabase entry with the greatest match score.
The system used to conduct this test split 1000 database
partitions between 14 server processes, two on each of
seven physical servers. Each physical server had two 1.0-
GHz Athlon processors and 2 Gbyte of RAM. Since each
partition is a different size, a simple, greedy method was
used to balance the total number of vectors assigned to
each server process, considering each partition in turn
from largest to smallest and assigning it to the server pro-
cess with the smallest total number of vectors so far. This
system computed matches for an entire recording against
its database of size 100 000 in approximately 1.5 s of CPU
time. This figure includes the client’s time spent generat-
ing query vectors from the audio input, the front-end da-
tabase server’s time spent dispatching queries, and the
back-end servers’ time spent matching the queries.
The entropy of the partitioned database was 8.45,
equivalent to a collection of 349 equally likely classifica-
tions. Fig. 9 shows the linear relationship, within this test
database, between the distance of a query vector from its
match target and the fraction of queries lost to partition
errors. In our test system, about 25% of queries were lost
to partition errors. However, this high error rate is well
tolerated by Tuneprint’s system of multiple independent
373
SCHMIDT AND BELMONTE
comparisons: as long as there are enough successful que-
ries to raise the score of the matching database entry above
those of nonmatching entries, the input will be correctly
identified. An increase in partition error simply produces a
corresponding increase in the length of input required for
reliable identification. For example, 150 fragments at 50%
loss will produce identifications as reliable as 100 frag-
ments at 25% loss, since on average both scenarios yield
the same number of successful queries, namely, 75. Al-
though the relationship between match scores and identi-
fication rate has not yet been systematically explored, our
experience is that the top match can be reliably detected if
it exceeds the next closest match by a score of 8 or more.
(Since matches at individual queries have values between
1/2 and 1, this threshold corresponds to matching at 8 to 16
individual fragments. Recall that the frames from which
each fragment is produced are spaced at intervals of
92.8575 ms. Therefore this threshold corresponds to an
input length of between about 0.75 and 1.5 s, neglecting
the effect of temporal autocorrelation between frames.
These figures match our experiences with inputs consist-
ing of short clips, where identification becomes unreliable
for input lengths below about 1 s.)
3 FURTHER WORK
The current Tuneprint system is a work in progress,
several aspects of which deserve further research. Some of
these issues are points that will become important as the
system scales up beyond the current 100 000 recordings.
Others are possibilities for improvement regardless of
scale.
With large databases, the question of what constitutes
an acceptable test set becomes increasingly important.
Tuneprint is meant to be applied to recordings that have
Fig. 9. Rate of query misclassification as a function of distance between query and target. Note linearity of relationship.
374
ENGINEERING REPORTS
undergone lossy compression and other subtle distortions
as are commonly found on file-sharing networks, and is
designed to implement an identity criterion of equivalence
to a human listener. Any validation of Tuneprint’s output
must therefore depend on a sampling of recordings repre-
sentative of those found on file-sharing networks, and on
the independent evaluations of human listeners which, for
this purpose of validation, cannot be automated. This re-
quirement for listening tests makes expansion of the test
set very labor intensive and is the reason for the current
test set’s small size. More significant than the size of the
test set, though, is the size of the database against which
the test recordings are being matched. It is the database
size that determines the system’s liability to collisions, and
hence to false positives and to match failures. Since the
elements of the test set are randomly selected, each ele-
ment constitutes an independent test of the database. Al-
though the further development of Tuneprint would ben-
efit from more detailed and larger scale testing than was
possible during the period of Tuneprint’s commercial
funding, the present result with a database of size 100 000
is an indicator of the promise of Tuneprint’s methods.
Partly because of the small size of Tuneprint’s test set,
a detailed, quantitative comparison with the performance
of other systems remains an open question. Wang [17]
independently implemented a temporal consistency mea-
sure similar to Tuneprint’s, though without full psycho-
acoustic modeling at the input stage; the experimental per-
formance of this system has not been reported. Using the
vector quantization method of Allamanche et al. [9], Hell-
muth et al. have implemented an advanced classifier [10]
that makes use of the information inherent in the temporal
sequence of feature vectors, though the exact nature of this
method is left unspecified. An alternative to vector quan-
tization methods is robust hashing, in which feature vec-
J. Audio Eng. Soc., Vol. 52, No. 4, April 2004
ENGINEERING REPORTS AUDIO IDENTIFICATION BY MULTIPLE INDEPENDENT PSYCHOACOUSTIC MATCHING
tors are discretized by simple thresholding of their com-
ponents, and the search space for approximate matches is
then constrained to only those database recordings of
which one or more frames produced an exact match in this
discretized domain [11]. Although the distribution of
within-recordings hash errors in a database of size 10 000
using this method theoretically predicts a very low rate of
false positives with a high rate of identification [12], the
between-recordings error rate has yet to be assayed ex-
perimentally in a large database. An important feature of
systems based on multiple independent comparisons is
that the method of analysis within time points is separable
from the method of evaluation for consistency across time
points. This separability raises the possibility of plugging
any within-points method into an alternative across-points
method—for example, Tuneprint’s psychoacoustically
based model could be integrated with a robust hashing
system.
Another key question regarding scaling is how the
growth of the database may increase the likelihood of
collision, that is, a situation in which two items in the
database map so close together in Tuneprint’s search space
that discrimination between them is degraded or impos-
sible. As Tuneprint matches first at the level of isolated
fragments and then at the level of whole recordings, there
are two senses in which collision can be considered. First,
collisions might occur between individual fragments of
different recordings. Such collisions would affect matching
of the recording as a whole in the same way that parti-
tioning error does. Tuneprint’s redundant strategy of mul-
tiple independent matching makes it robust to collisions at
the fragment level in the same way that it is robust to
partitioning error—as long as there are significantly many
successful matches, failed matches do not affect the iden-
tification [15]–[17]. Although in tests to date any small
effect of such collisions has been swamped by the effect of
partitioning error, a high rate of collisions could be ex-
pected to mimic the effect of partitioning error, increasing
the length of input required for reliable identification.
Second, one could imagine a collision involving very
strong matching to more than one recording in the data-
base. Although we have observed such occurrences during
our tests, all have turned out to involve cases in which a
recording loaded into the database from a peer-to-peer
network had been mislabeled and was actually a copy of
another matching recording already present in the data-
base. We have never observed a true collision at the level
of whole recordings, and this absence of collision may
perhaps be expected given the length of a typical recording
and the number of dimensions within which it can vary as
it evolves through time.
Scaling certainly can be expected to figure into the
tradeoff between entropy and partitioning error. It remains
undetermined what degree of partitioning would be opti-
mal even at the database’s current size. The limit in which
partitioning errors begin to affect accuracy has not yet
been reached, and the current stopping point for the num-
ber of partitions is somewhat arbitrary. Assuming that the
rate of partitioning errors remains a linear function of the
distance between the query and its best match (Fig. 9), a
J. Audio Eng. Soc., Vol. 52, No. 4, April 2004
database partitioning can be characterized by the slope of
this partitioning-error function in combination with the
entropy figure. The total expected CPU time per identifi-
cation is the product of the expected number of queries (a
function of the partitioning-error rate) and the expected
CPU time per query (a function of entropy). It likely will
be possible to develop an optimization procedure that ap-
plies these descriptive statistics to find the partitioning
scheme that minimizes the expected CPU time per
identification.
Perhaps the most compelling question regarding parti-
tioning is that of how much information needs to be pre-
served within a single partition. Vector quantization yields
an efficiency of time, arising from the restriction of the
search space to a particular partition, and an efficiency of
space, arising from the lossy coding of individual vectors
in terms of the partition to which each maps. Currently
Tuneprint takes advantage only of the temporal efficiency,
applying vector quantization to select a particular database
partition to search. The potential spatial efficiency is not
realized, since the original, unquantized vectors are pre-
served for use in nearest-neighbor matching within the
selected partition. A database with very high entropy
might offer the potential for eliminating the query-to-
match distance measure within a partition, and instead
treating all elements of the partition as equally good
matches. The winning match at the level of whole record-
ings could then be determined by consistency of matching
as in the current model. Such an ability to discard the
original vectors would offer large savings in space, by
eliminating the bulk of the database, and in time, by elimi-
nating the demand for nearest-neighbor matching in a
high-dimensional space. We have implemented a proto-
type of such a system which compresses the entire data-
base into 3 Gbyte, runs on a single CPU, and, when suc-
cessful, identifies a recording in only 110 ms of processing
time on average.
Variations on this high-entropy strategy also are pos-
sible. One optimization might involve overlapping parti-
tions, that is, allowing database vectors that lie near par-
tition boundaries to be included in more than one partition.
Conversely, query vectors that lie near partition bound-
aries could be made to trigger a search of multiple parti-
tions in the database. Either of these strategies would de-
crease the rate of partitioning error, at the expense of a
modest increase in search time. One can also imagine a
two-pass system in which distance measures are computed
only for those match candidates that show consistent
matching over time at the level of partitions.
Currently the query is sampled at constant intervals
throughout its length. Higher confidence in matching
likely can be achieved by varying this sampling period,
both as a function of overall match confidence (more sam-
pling of difficult-to-identify recordings) and as a function
of the temporal derivative of the psychoacoustic function
(more sampling in the time intervals surrounding abrupt
changes). Landmarking of the input recording is one way
of implementing increased sampling at intervals of abrupt
change, and it would be interesting to compare the perfor-
mance of landmarking based on simple acoustic properties
375
SCHMIDT AND BELMONTE
[17] with landmarking based on psychoacoustic proper-
ties. Although in general the psychoacoustic output varies
slowly over time, the presence of two principal compo-
nents that encode time-varying information from neigh-
boring fragments is an indication that abrupt changes con-
tain much useful information.
Although dimensionality reduction by principal compo-
nents analysis is an expedient strategy, it may not be the
best way to form a basis. In particular, the roles of the two
time-varying components extracted by PCA have yet to be
ascertained. In an environment in which queries may be
offset from their best matches by up to half the length of
a fragment, an inclusion of time-varying components may
simply add noise. In addition, the construction of a space
based on independent component analysis (ICA) [26] has
not yet been explored, and should be.
Although Tuneprint has performed well in preliminary
tests, the effects of distortion on Tuneprint’s performance
have not yet been explored systematically. Since our
test set was selected at random from recordings publicly
available on peer-to-peer networks, its audio quality was
representative of this population. Our statistics indicate
that only about 1% of these recordings are encoded at
bit rates less than 128 kbits/s. In addition, our bandwidth
of 253 to 12 500 Hz ranges four times as high as that
available in telephone transmission. Testing and retuning
for low-bit-rate encodings, for signals varying in playback
speed, and for signals varying in bandwidth and other
equalization properties were under way with positive
early results, but unfortunately could not be completed
within the period of Tuneprint’s commercial funding.
The outlook for low-bandwidth processing seems par-
ticularly positive since most of the amplitude of the
basis vectors (see Fig. 4) is contained in the lower Bark
frequencies.
It is worth noting that audio is not the only medium to
which our general strategy of multiple independent match-
ing of fragments might apply. Any medium that can be
reduced to an array of features, either over time or over
spatial dimensions, can be subjected to fragmentation over
these dimensions. A fragment analysis of images, for in-
stance, might compare features that are independent of
translation, rotation, and scaling, and the corresponding
assembly function might compute affine transformations
between the query image and matching portions of the
database images.
We have described a system for content-based identifi-
cation of audio recordings, built around a strategy of mul-
tiple independent matching. Using a psychoacoustically
based representation, this system matches inputs according
to a criterion meant to model that of a human listener. Its
identifications are robust to truncation, and work well with
the lossy compression commonly found in recordings on
peer-to-peer networks. In addition, the system is designed
to be robust to modest variations in equalization and play-
back speed. For equalization in particular, we found that
high-pass filtering of the output of the psychoacoustic
transform (see Fig. 3) was very effective in maintaining
performance. Successful operation on a database of
100 000 recordings bodes well for further scaling.
376
ENGINEERING REPORTS
4 ACKNOWLEDGMENT
Profound thanks are due to our colleagues Daren Gill,
Martin Stiaszny, Amittai Axelrod, Josh Pollack, Jennifer
Chung, and Lex Nemzer, without whose many day and
night hours the Tuneprint system could never have been
implemented. In addition, we wish to acknowledge Sage
Hill Partners, who funded the development of Tuneprint.
5 REFERENCES
[1] A. Pikrakis, S. Theodoridis, and D. Kamarotos,
“Recognition of Isolated Musical Patterns in the Context
of Greek Traditional Music,” presented at the IEEE Int.
Conf. on Electronics, Circuits, and Systems, Rhodes,
Greece, 1996 Oct. 13–16.
[2] J. T. Foote, “Content-Based Retrieval of Music and
Audio,” in C. C. J. Kuo, Ed., Multimedia Storage and Ar-
chiving Systems II, Proc. SPIE, vol. 3229, pp. 138–147
(1997).
[3] A. Pikrakis, S. Theodoridis, and D. Kamarotos,
“Recognition of Isolated Musical Patterns Using Discrete
Observation Hidden Markov Models,” presented at the
European Signal Processing Conf., Rhodes, Greece, 1998
Sept. 8–11.
[4] D. Pye, “Content-Based Methods for the Manage-
ment of Digital Music,” in Proc. IEEE Int. Conf. on
Acoustics, Speech, and Signal Processing (Istanbul, Tur-
key, 2000), vol. 4, pp. 2437–2440.
[5] C. Papaodysseus, G. Roussopoulos, D. Fragoulis,
T. Panagopoulos, and C. Alexiou, “A New Approach to
the Automatic Recognition of Musical Recordings,” J. Au-
dio Eng. Soc., vol. 49, pp. 23–35 (2001 Jan./Feb.).
[6] C. J. C. Burges, J. C. Platt, and S. Jana, “Distortion
Discriminant Analysis for Audio Fingerprinting,” IEEE
Trans. Speech Audio Process., vol. 11, pp. 165–174
(2003).
[7] S. Sukittanon and L. E. Atlas, “Modulation Fre-
quency Features for Audio Fingerprinting,” presented at
the IEEE Int. Conf. on Acoustics, Speech, and Signal Pro-
cessing, Orlando, FL, 2002 May 13–17.
[8] E. Wold, T. Blum, D. Keislar, and J. Wheaton,
“Content-Based Classification, Search, and Retrieval of
Audio,” IEEE Multimedia, vol. 3, pp. 27–36 (1996).
[9] E. Allamanche, J. Herre, O. Hellmuth, B. Fröba,
and M. Cremer, “AudioID: Towards Content-Based Iden-
tification of Audio Material,” presented at the 110th Con-
vention of the Audio Engineering Society, J. Audio Eng.
Soc. (Abstracts), vol. 49, p. 542 (2001 June), convention
paper 5380.
[10] O. Hellmuth, E. Allamanche, J. Herre, T. Kastner,
M. Cremer, and W. Hirsch, “Advanced Audio Identifica-
tion Using MPEG-7 Content Description,” presented at the
111th Convention of the Audio Engineering Society, J.
Audio Eng. Soc. (Abstracts), vol. 49, pp. 1227–1228 (2001
Dec.), convention paper 5463.
[11] J. Haitsma, T. Kalker, J. Oostveen, “Robust Audio
Hashing for Content Identification,” presented at the In-
ternational Workshop on Content-Based Multimedia In-
dexing, Brescia, Italy, 2001 Sept. 19–21.
J. Audio Eng. Soc., Vol. 52, No. 4, April 2004
ENGINEERING REPORTS AUDIO IDENTIFICATION BY MULTIPLE INDEPENDENT PSYCHOACOUSTIC MATCHING
[12] J. Haitsma and T. Kalker, “A Highly Robust Audio
Fingerprinting System,” presented at the 3rd Int. Conf. on
Music Information Retrieval, Paris, France, 2002 Oct. 13–17.
[13] R. J. McNab, L. A. Smith, I. H. Witten, C. L. Hen-
derson, and S. J. Cunningham, “Towards the Digital Mu-
sic Library: Tune Retrieval from Acoustic Input,” in Proc.
1st ACM Int. Conf. on Digital Libraries (Bethesda, MD,
1996, pp. 11–18.
[14] J. J. Hopfield and C. D. Brody, “What Is a Mo-
ment? Transient Synchrony as a Collective Mechanism for
Spatiotemporal Integration,” Proc. Nat. Acad. Sci. USA,
vol. 98, pp. 1282–1287 (2000).
[15] G. Schmidt, “Tuneprint Audio Fingerprinting
Technology: Technical Overview Version 1.1,” paper pre-
pared in response to Recording Industry Association of
America/IFPI request for information (2001 June 6).
[16] G. Schmidt, “Tuneprint Fingerprinting Technol-
ogy,” presented to the Recording Industry Association of
America, Washington, DC (2001 Sept. 5).
[17] A. Wang, “System and Methods for Recognizing
Sound and Music Signals in High Noise and Distortion,”
International Patent Publ. WO 02/11123 A2 (2002).
[18] E. Zwicker and H. Fastl, Psycho-acoustics: Facts
and Models, 2nd ed. (Springer, Berlin, 1999).
[19] K. Brandenburg and G. Stoll, “ISO/MPEG-1 Au-
dio: A Generic Standard for Coding of High-Quality Digi-
THE AUTHORS
G. Schmidt
Geoff Schmidt was born in Branson, MO, in 1980. He
left undergraduate studies at the Massachusetts Institute of
Technology in Cambridge, MA, after one term to pursue
entrepreneurial interests. Working alone, he developed the
Tuneprint framework in mid-2000 and in 2001 secured
venture capital funding from Sage Hill Partners in Cam-
bridge to scale up and commercialize the system. Tune-
print Corporation ceased operations in 2002.
Mr. Schmidt’s previous work on output-sensitive vis-
ible surface determination algorithms for 3-D rendering
has won awards from the International Science and Engi-
neering Fair, the U.S. Junior Science and Humanities Sym-
posium, and other research competitions. He is now a com-
mercial research consultant, currently performing machine
vision research at Intellivid, a Cambridge startup company.
Mr. Schmidt’s personal interests include meditation,
teaching, and politics. He recently served as campaign
manager for Matt DeBergalis’s Cambridge City Council
campaign, where in addition to day-to-day management he
designed a successful direct mail effort. He plans to pursue
J. Audio Eng. Soc., Vol. 52, No. 4, April 2004
tal Audio,” J. Audio Eng. Soc., vol. 42, pp. 780–792 (1994
Oct.).
[20] J. G. Beerends and J. A. Stemerdink, “A Percep-
tual Audio Quality Measure Based on a Psychoacoustic
Sound Representation,” J. Audio Eng. Soc., vol. 40, pp.
963–978 (1992 Dec.).
[21] ISO 389-7 “Acoustics—Reference Zero for the
Calibration of Audiometric Equipment—Part 7: Reference
Threshold of Hearing under Free-Field and Diffuse-Field
Listening Conditions,” International Organization for
Standarization, Geneva, Switzerland (1996).
[22] B. C. J. Moore, R. W. Peters, and B. R. Glasberg,
“Detection of Decrements and Increments in Sinusoids at
High Overall Levels,” J. Acoust. Soc. Am., vol. 99, pp.
3669–3677 (1996).
[23] I. T. Jolliffe, Principal Component Analysis, 2nd
ed. (Springer, New York, 2002).
[24] Y. Linde, A. Buzo, and R. M. Gray, “An Algo-
rithm for Vector Quantizer Design,” IEEE Trans. Com-
mun., vol. 28, pp. 84–95 (1980).
[25] S. P. Lloyd, “Least Squares Quantization in PCM,”
IEEE Trans. Inform. Theory, vol. 28, pp. 129–137 (1982).
[26] A. J. Bell and T. J. Sejnowski, “An Information-
Maximization Approach to Blind Separation and Blind
Deconvolution,” Neural Comput., vol. 7, pp. 1004–1034
(1995).
a career in machine learning or programming language
design. E-mail gschmidt@mit.edu.
●
Matthew Belmonte studied English literature and com-
puter science as an undergraduate, developing an interest
in the processing of formal and natural languages. Moving
from artificial to biological computing systems, he applied
this computational interest to problems in cognitive neu-
roscience. He is the author of several papers on the neu-
rophysiology of attention and perception in normal and
autistic populations and on computational methods for sta-
tistical analysis of biophysical time series. He was a re-
search scientist at Tuneprint Corporation, the architect of
Tuneprint’s psychoacoustic model, and a major contribu-
tor to the technical description of the Tuneprint system. He
left the United States in 2002 and currently works in func-
tional magnetic resonance imaging at the University of
Cambridge Autism Research Centre in the UK. E-mail:
belmonte@mit.edu.
377
 ENGINEERING REPORTS

On the Detection of Melodic Pitch in a

Percussive Background*

PREETI RAO, AES Member, AND SAURABH SHANDILYA

Department of Electrical Engineering, Indian Institute of Technology, Bombay, India

The extraction of pitch (or fundamental frequency) information from polyphonic audio
signals remains a challenging problem. The specific case of detecting the pitch of a melodic
instrument playing in a percussive background is presented. Time-domain pitch detection
algorithms based on a temporal autocorrelation model, including the Meddis–Hewitt algo-
rithm, are considered. The temporal and spectral characteristics of percussive interference
degrade the performance of the pitch detection algorithms to various extents. From an
experimental study of the pitch estimation errors obtained on a set of synthetic musical
signals, the effectiveness of the auditory-perception–based modules of the Meddis–Hewitt
pitch detection algorithm in improving the robustness of fundamental frequency tracking in
the presence of percussive interference is discussed.

0 INTRODUCTION Pitch determination of speech signals has been the sub-

The problem of pitch (or fundamental frequency) ex-
traction of periodic signals in the presence of interfering
sounds and noise is an important problem in both speech
and music applications. Apart from the value of pitch in-
formation per se, a knowledge of the time-varying funda-
mental frequency can be useful in the separation and re-
construction of a harmonic source from a sound mixture.
A number of pitch detection algorithms (PDAs) have been
proposed over the decades. But while each has had a mea-
sure of success in the targeted application, no single PDA
is found suitable for all types of signals and conditions.
This engineering report presents an investigation of the
performance of some well-known PDAs in estimating the
fundamental frequency of a melodic instrument playing in
the presence of percussive background. This is a restricted
case of the larger problem of musical pitch detection in
polyphony. Nevertheless it is an important problem. For
instance, classical Indian vocal and instrumental music is
always accompanied by percussive instruments providing
the rhythmic structure. The melody itself is strongly char-
acterized by the presence of microtones and continuous
pitch variation. Detecting the melodic pitch contour has
important applications in music recognition and for gen-
erating metadata in audio content retrieval systems.
*Manuscript received 2003 July 9; revised 2004 February 9.
An early version of this work was presented at the 114th Con-
vention of the Audio Engineering Society, Amsterdam, The
Netherlands, 2003 March 22–25, under the title “Pitch Detection
of the Singing Voice in Musical Audio.”
378
ject of research for decades [1]. It is marked by challenges
arising from the complex temporal and spectral structure
of speech as well as its nonstationary nature. While mu-
sical applications require a higher accuracy of pitch esti-
mation than speech applications, tracking the pitch of a
melodic instrument or singing voice is easier than tracking
that of speech signals due to the relatively slowly changing
signal characteristics. The presence of interfering sounds
from percussive accompaniment, however, would be ex-
pected to adversely affect the accuracy of pitch estimates
of any given PDA for musical applications. Percussive
sounds are characterized by rapidly varying temporal en-
velopes, mixed partials-plus-broad-band noise spectra, and
low values of signal-to-interference ratio in localized time
intervals. The peculiar problems in pitch detection posed
by such interference form the main focus of the present
study. In particular, we consider the degradation caused by
the presence of inharmonic interfering partials. The ro-
bustness of pitch detection methods to additive broad-band
noise has been studied in various contexts in the literature
(see, for example, [2]). While the motivation for the pres-
ent work is the pitch tracking of the melodic instruments
including the singing voice in the presence of percussion,
we use in our experiments test signals from a set of MIDI
instrument voices to enable controlled experiments focus-
ing on the effects of percussion, with access to the “ground
truth” pitch available.
In this study an important subclass of PDAs, namely,
those based on the detection of periodicity in the time-
domain signal by means of short-term correlation, is con-
sidered. Autocorrelation-based pitch determination, used
J. Audio Eng. Soc., Vol. 52, No. 4, April 2004
ENGINEERING REPORTS
widely in speech analysis [1], has also been found suitable
for the pitch tracking of monophonic musical signals [3].
The present study investigates the Meddis–Hewitt percep-
tual PDA as an example of a more sophisticated algorithm
also based on the detection of periodicity via temporal
autocorrelation.
The engineering report is organized as follows. Section
1 provides a brief overview of various PDAs, with an
introduction to the PDAs chosen for the present study. The
subsequent sections describe the implementation of the
functional blocks of the PDAs and the evaluation of
the PDAs by an experiment on the synthetic signal test
set. The study concludes with a discussion of the obser-
vations targeted toward obtaining insights into the perfor-
mance of the PDAs with respect to signal and interference
characteristics.
1 PITCH DETECTION ALGORITHMS
Time-domain PDAs, the oldest pitch detection algo-
rithms, are based on measuring the periodicity of the sig-
nal via the repetition rate of specific temporal features.
Frequency-domain PDAs, on the other hand, are based on
detecting the harmonic structure of the spectrum. Among
the simpler time-domain PDAs is the popular autocorre-
lation function (ACF)–based PDA. The definition of the
“biased” autocorrelation function is given by [1]
N−␶−1
ACF共k,␶兲 = 兺 y共k + i)y(k + i + ␶)
i=0
(1)
where k and ␶ are position of window and correlation lag,
respectively, and y is the input signal.
For a pure tone, the ACF exhibits peaks at lags corre-
sponding to the period and its integral multiples. The peak
in the ACF of Eq. (1) at the lag ␶ corresponding to the
signal period will be higher than that at the lag values
corresponding to multiples of the period. For a musical
tone consisting of the fundamental frequency component
and several harmonics, one of the peaks due to each of the
higher harmonics occurs at the same lag position as that
corresponding to the fundamental, in addition to several
other integer multiples of the period (subharmonics) of
each harmonic. Thus a large peak corresponding to the
sum contribution of all spectral components occurs at the
period of the fundamental (and higher integral multiples of
the period of the fundamental). This property of the ACF
makes it very suitable for the pitch tracking of monopho-
nic musical signals. The ACF PDA chooses as the pitch
period the lag corresponding to the highest peak within a
range of lags.
In contrast to the simplicity of the ACF pitch detector
are more recent PDAs, also based on autocorrelation, but
derived more closely from the mechanism of temporal
coding in the human auditory system. These PDAs can in
fact be viewed in terms of preprocessing of the signal
followed by autocorrelation-based detection. There are a
number of variants in this class of PDAs, but they all share
some important characteristics [4]. They decompose the
signal into frequency bands defined by the auditory filters
of the cochlea. Next, nonlinear processing corresponding
J. Audio Eng. Soc., Vol. 52, No. 4, April 2004
DETECTION OF MELODIC PITCH IN A PERCUSSIVE BACKGROUND
to hair-cell transduction is applied and the temporal peri-
odicity detected separately in each frequency channel by
means of autocorrelation. Finally the across-channel infor-
mation is combined to produce a single pitch estimate. The
recent pitch perception model of Meddis and Hewitt [5]
has gained much prominence due to its demonstrated abil-
ity to predict the results of certain crucial pitch perception
experiments. The Meddis–Hewitt PDA is based on the
functional modules of the auditory periphery with added
processing stages that emulate auditory processing, which
is considered to be more central. The various stages of the
PDA are a bandpass filter representing the transfer func-
tion of the outer ear and middle ear canal, a bank of filters
modeling the basilar membrane response, followed by a
model of the inner hair cell applied to each filter channel
output to simulate neural transduction, obtaining a series
of firing probabilities. Next an ACF periodicity detector is
applied to each of the hair-cell model outputs. Finally a
summary autocorrelation function (SACF) is formed by
adding the ACFs so obtained across the frequency chan-
nels. A search for the highest peak in the relevant range of
SACF lags provides an estimate of the pitch period.
The added presence of noise and inharmonic partials
due to an interfering signal perturbs the shape and location
of the peaks contributed by the signal harmonics in the
ACF. Thus the traditional ACF pitch detector applied to a
musical signal with percussive accompaniment would be
expected to be adversely affected by the presence of noise
and inharmonic frequency components contributed by the
percussion. On the other hand, in the case of perception-
based PDAs, the signal is processed by a number of au-
ditory-model–based blocks before being subjected to pe-
riodicity extraction via the ACF. In particular, a combina-
tion of linear and nonlinear filtering is applied, and the
temporal periodicity information itself is computed via the
ACF separately in each frequency channel. It is of interest
to examine whether and to what extent these perceptually
motivated enhancements improve the reliability of the
pitch estimation for harmonic musical signals with con-
spicuous background interference. While much recent
work has investigated the ability of perceptual PDAs to
predict subjectively perceived pitch in psychoacoustic ex-
periments, the present work examines the robustness of the
PDA for estimating the signal fundamental frequency in
the presence of interference. By means of carefully de-
signed test signals, we study the pitch estimation errors
obtained by the PDAs in the presence of percussive inter-
ference with respect to the underlying pitch of the melodic
voice, and later attempt to explain these observations.
2 IMPLEMENTATION OF PDA
FUNCTIONAL BLOCKS
Fig. 1 provides a modular structure for the PDA based
on the Meddis–Hewitt pitch perception model [5]. The
individual blocks represent various stages of the algo-
rithm, each of which may, in principle, be implemented in
multiple ways.
Block 1 represents the outer ear and middle ear (OEM)
prefiltering, with the magnitude response shown in Fig. 2.
379
RAO AND SHANDILYA
Essentially a bandpass filter with a resonance frequency
near 3 kHz, this block has been implemented by the cas-
cade of an eighth-order low-pass IIR filter and a second-
order parameterized high-pass filter with a high-pass filter
parameter value of 0.94 [6]. The magnitude response is
similar to the inverted absolute threshold-of-hearing curve,
whereas below 1 kHz it is approximated by the inverted
equal-loudness contour for high loudness levels.
Block 2 represents the cochlear filter bank with filter
center frequencies that are equally spaced on the ERB
(equivalent rectangular bandwidth) scale with the band-
width increasing with the center frequency. A filter bank
of 27 ninth-order gammatone filters of bandwidth 1 ERB
each are based on the corresponding function of the
HUTear library [7]. These filters have center frequencies
ranging from 123 Hz to 5.636 kHz, or 4 to 30 on the ERB
scale. The output of each filter simulates the pattern of
vibration at a particular location on the basilar membrane.
Next the conversion of this mechanical activity to the
neural spike generation events of the auditory nerve is
simulated by block 3. The implementation of this module
Fig. 1. Block diagram of functional blocks of Meddis–Hewitt PDA with postprocessing added.
Fig. 2. Magnitude spectrum of outer-ear–middle-ear (OEM) filter.
380
ENGINEERING REPORTS
can range from a full model of the hair cell derived from
a computational analysis of actual hair-cell and auditory-
nerve processes [5] to simple half-wave rectification fol-
lowed by low-pass filtering [8]. In 1986 Meddis proposed
a model for the hair cell which simulates, through differ-
ence equations rich in parameters, several properties of the
neural transduction and auditory nerve firing [9]. Impor-
tant characteristics of the hair-cell model are its nonlin-
earity and frequency selectivity. The present implementa-
tion is based on the hair-cell model of Meddis and
coworkers [9], [10] as implemented in the auditory model
library of Slaney [11].
Block 4 calculates the ACF [Eq. (1)] of the signal input
to this block. Keeping in mind the range of 150–800 Hz
for the expected fundamental frequency, we have used a
40-ms window with 50% overlap (frame space 20 ms) for
the computation of the ACF. Once the ACFs are obtained
for all the channels [by implementing Eq. (1) on the hair-
cell model output of each channel], block 5 performs the
task of combining them. Combining can occur in the form
of either simple or weighted addition. We use simple sum-
J. Audio Eng. Soc., Vol. 52, No. 4, April 2004
ENGINEERING REPORTS
ming. The combined ACFs are known as summary ACF
(SACF). In block 6 the SACF is searched for the highest
peak within a prespecified range (corresponding to the
expected fundamental frequency range of 150–800 Hz).
The lag value corresponding to the highest peak is ac-
cepted as the estimated pitch period. Block 7 is a postpro-
cessing block, which smoothens out local variations in the
pitch estimates across frames using a simple three-point
median filter. The combination of all seven blocks consti-
tutes the Meddis–Hewitt PDA with added postprocessing.
In the next section we describe a procedure for evaluating
the contribution of the various functional blocks of the
Meddis–Hewitt PDA to improving robustness over direct
ACF peak-based pitch detection.
3 EXPERIMENTAL EVALUATION
In order to investigate the performance of the pitch de-
tection algorithms for the pitch estimation of a harmonic
signal in a percussive background, a set of test signals was
designed from available MIDI songs. Apart from the ac-
cess to the “ground truth” pitch, the use of MIDI files
provides great flexibility by allowing the inclusion or
elimination of individual monophonic instrument chan-
nels, modifying the relative strengths of the component
sounds, pitch transformations, and choice of the instru-
ment playing the melody line as well as the percussive
instrument.
3.1 Test Signal Set
A MIDI song of length 8 seconds was selected. It had a
single harmonic instrument (alto sax) playing the melody
accompanied by several percussive (nonpitched) voices,
namely, hi hat, kick drum, and low agogo channels. The
pitch range of the melody was 350–620 Hz and consisted
of four similar phrases, with each phrase comprising five
notes of various durations. [The pitch contour of the
melody can be seen as the solid line in Fig. 5(b).] Further,
Fig. 3. Spectrogram illustrating time–frequency behavior of three percussion instruments.
J. Audio Eng. Soc., Vol. 52, No. 4, April 2004
DETECTION OF MELODIC PITCH IN A PERCUSSIVE BACKGROUND
in order to create a number of different test conditions,
pitch-shifted versions of the melody were created as fol-
lows: high (up by 4 semitones to the range of 440–787 Hz)
or low (down by 12 semitones to the range of 174.5–311
Hz). It may be remarked that the pitch transformations are
achieved via “instrumental” pitch shifting, which implies
that the relative amplitudes of the harmonics remain un-
changed across fundamental frequency changes, in con-
trast to formant-corrected pitch shifting.
The set of percussive instruments represents a range of
signal characteristics, as illustrated in the spectrograms of
Fig. 3. The kick drum is a relatively fast decaying signal
with predominantly low-frequency content. The hi hat is
characterized by a slow time decay and a broadly spread
spectral mixture of moderately strong partials and noise.
Low agogo has low-noise content and strong partials all
the way from 1.1 to 10 kHz, with a moderate rate of decay.
To obtain a variety of combinations of target and inter-
ference timbres, the song was transformed by changing the
target instrument and then selecting only one of the inter-
fering (percussive) instruments at a time. The selected
target instrument voices were of different timbres, as
shown by the magnitude spectra of a fixed note in Fig. 4.
For the middle- and high-frequency ranges, baritone sax
(prominent high harmonics), flute (weak high harmonics),
and oboe (harmonics spread in frequency) were used. For
the low-frequency range the flute was replaced by alto sax
to incorporate a more natural sound.
The JAZZ MIDI sequencer, available as shareware,1
was used to achieve the needed transformations. The
melody line was recorded, switching off all the other chan-
nels, in each of the three pitch ranges with each of three
target instruments. Thus a set of nine files containing pure
melody was obtained. For each of these melody files we
created three “corrupted” versions, each with only a single
percussion channel turned on. The synchronization be-
1
www.jazzware.com.
381
RAO AND SHANDILYA
tween the melody and each of the percussion tracks was
such that a minimum number of percussion strikes fell in
the silence region between target instrument notes. This
led to percussion onsets being located at a variety of po-
sitions with respect to target note onset, steady state, and
decay. Then relative amplitudes of the target and interfer-
ing signals were set so that the ratio of signal power to
interference power (each of the powers being computed as
the corresponding average over the nonsilent regions of
the musical piece) remained at a fixed predefined value
(equal to 2.0) for a set of test signals across various target
instruments, percussive instruments, and pitch ranges. The
signal-to-interference ratio, however, is only an average
value with local values deviating greatly, depending on the
position of the percussion strike with respect to the target
instrument note onset. We thus obtained a total of 27 test
signals, all sampled at 44.1 kHz.
3.2 Experiment
The PDAs were run on the pure test signals, with the
same postprocessing applied to each PDA estimate to en-
sure a fair comparison.
Fig. 5 shows a sample of the experimental results ob-
tained. The selected test signal is the combination of bari-
tone sax and low agogo percussion. Fig. 5(a) shows the
spectrogram of the test signal. The relatively continuous
dark lines correspond to the harmonic partials of the bari-
tone sax, and the very short dark segments that occur
during the first (2 strikes), third, and fourth notes of each
phrase correspond to the partials of the low agogo. Fig.
5(b) compares the true pitch track with the pitch track
estimated by the traditional ACF pitch detector followed
by a three-point median filter. We see that there are large
pitch estimation errors, which coincide with the occur-
rence of the percussion and last over several frames. The
Fig. 4. Magnitude spectra of melodic instruments used in experimental study.
382
ENGINEERING REPORTS
simple three-point median postfilter only corrects isolated
pitch errors. The instances of percussion where pitch er-
rors do not occur seem to be characterized by overlapping
partials between target and percussion. Fig. 5(c) illustrates
the pitch contour obtained by using the Meddis–Hewitt PDA,
also followed by the postfilter, on the same test signal. Indeed
the performance has improved on using this algorithm.
A controlled-parameter experiment was carried out to
study the behavior of the PDAs on the underlying signal
characteristics, and to obtain an understanding of the role
of each of the functional modules of the Meddis–Hewitt
algorithm in influencing the pitch detection. Specifically,
four PDAs were evaluated on the test data set. All four
algorithms are derived from the generic block diagram of
Fig. 1 by choosing different combinations of subblocks
and/or different realizations of a specific subblock. The
postprocessor of block 7 is a three-point median filter that
is included in all four algorithms. The details of the four
PDAs are as follows.
1) AC1: This algorithm incorporates blocks 4, 6, and 7
only. It corresponds to the traditional ACF pitch detector.
Block 4, that is, the ACF calculation block, uses a rectan-
gular window and the biased ACF computation of Eq. (1).
2) AC2: This algorithm incorporates block 1 with
blocks 4, 6, and 7. This again is a traditional ACF pitch
detector, but with outer ear/middle ear filtering included as
a preprocessing function.
3) AC3: This algorithm comprises blocks 1, 3, 4, 6,
and 7. This algorithm is an extension of AC2, where the
neural transduction block (based on the hair-cell model of
[9]) has been introduced. Unlike the Meddis–Hewitt PDA,
the signal is not decomposed into separate frequency chan-
nels. Rather, the hair-cell nonlinearity is applied to the full
band signal followed by ACF pitch detection to result in a
single estimate of temporal periodicity per frame.
J. Audio Eng. Soc., Vol. 52, No. 4, April 2004
ENGINEERING REPORTS
4) MH1: All blocks 1 to 7 are included, making this a
complete implementation of the Meddis–Hewitt algorithm
with postprocessing added. For the ACF, a Hamming-
windowed biased ACF is computed for each channel using
an efficient fast Fourier transform implementation.
3.3 Observations
The bar charts of Figs. 6 to 9 display the results of the
experiment in terms of a count of the pitch errors with
respect to the known reference pitch contour, arranged by
the PDA configuration used. A pitch estimate is obtained
for every analysis frame only in regions where the target
instrument is playing. At a frame spacing of 20 ms, this
comprises a total of 285 frames. A pitch error is defined to
occur whenever the detected pitch deviates from the
reference pitch by 3% or more (about half a semitone) of
the reference pitch frequency. Of the detected pitch
errors, those of magnitude less than 6% are labeled “fine”
errors, whereas those of higher magnitude are labeled
“gross.” The gross errors are found typically to be pitch
octave errors. It may be noted that in the absence of per-
cussion, no pitch errors were observed in any of the PDA
configurations.
Fig. 5. Pitch estimates for test signal “middle pitch, baritone sax with low agogo.” (a) Input signal spectrogram (prominent low-
frequency partials of percussion encircled). (b) Actual pitch (——) and pitch estimated from AC1 PDA (- - -). (c) Actual pitch (——)
and pitch estimated from MH1 PDA (- - -).
J. Audio Eng. Soc., Vol. 52, No. 4, April 2004
DETECTION OF MELODIC PITCH IN A PERCUSSIVE BACKGROUND
From an inspection of the bar charts we note that the
extent of pitch errors depends not only on the PDA but
also on the percussion instrument, the pitch range of the
target instrument, and the target instrument itself. Of the
latter three factors, the most marked is the dependence of
the PDA performance on the percussion characteristics.
The ACF pitch detector AC1 makes a large number of
errors for all three percussions. The target instrument pitch
range that is most affected is seen to depend on the spec-
tral characteristics of the interference. The kick drum with
its low-frequency support affects the lowest pitch range
the most, whereas the low agogo with its broad spectral
mixture of partials and noise impacts all three pitch ranges.
The introduction of OEM filtering in AC2 has the effect
of an overall lowering of the extent of pitch errors. A
strong exception to this happens to be the hi hat in the
high-pitch range (note the changed scale of the error axis).
The hair-cell model followed by autocorrelation (AC3)
serves to reduce all errors further, with the only significant
errors remaining in the low agogo signals. Finally, the
full Meddis–Hewitt PDA MH1 reduces the errors in the
low agogo signals of the low- and middle-pitch ranges, but
worsens the performance slightly in the high-pitch range.
383
RAO AND SHANDILYA
4 DISCUSSION
In each of the PDA configurations used in the experi-
ment, the final pitch estimate (prior to postprocessing) is
obtained by searching the ACF (or SACF) for the lag
value corresponding to the highest peak within the ex-
pected range of lags. The ACF is computed either directly
on the input signal or after nonlinear processing by the
hair-cell model. In the case of the Meddis–Hewitt PDA,
the hair-cell nonlinearity followed by the ACF is com-
puted separately in frequency bands as determined by the
gammatone filter bank, and then combined linearly to ob-
tain the SACF. To understand the behavior of ACF peak-
based pitch detection better it is useful to think of the ACF
of a signal comprising several components (harmonic and
Fig. 6. Error performance of AC1 PDA for various target instruments and pitch ranges with percussion instrument in background. (a)
Kick drum. (b) Hi hat. (c) Low agogo.
384
ENGINEERING REPORTS
inharmonic) as the inverse Fourier transform of the power
spectrum of the signal [1]. The signal power spectrum is
insensitive to the relative phases of the components, to the
extent that the window is long enough that there is no
significant leakage of the frequency components. Due to
the linearity of the Fourier transform, the ACF of the
signal is the summation of the ACFs of the individual
components in the signal power spectrum, and is therefore
insensitive to the phase relations between components. Based
on this interpretation of the ACF, the observations of the
previous section are discussed and justified via simula-
tions using simplified implementations of channel separa-
tion (ideal bandpass filters) and hair-cell nonlinearity (a
half-wave rectifier followed by a low-pass filter given by
first-order Butterworth with 1-kHz cutoff frequency).
J. Audio Eng. Soc., Vol. 52, No. 4, April 2004
ENGINEERING REPORTS
For the pure target signals, each of which contains a
number of harmonics, including the fundamental, the win-
dowed ACF of the input signal computed according to Eq.
(1) shows peaks at lags corresponding to the pitch period
and multiples of the pitch period. The highest peak corre-
sponds to the pitch period, and there is no error in the
estimated pitch. On the other hand, when the input signal
to the ACF contains noise or interfering partials, there is a
perturbation of the peak corresponding to the correct pitch
period. The ACF of the interference partial (which can be
considered to combine additively with the target ACF to
form the corrupted ACF) modifies the values of the origi-
nal ACF at all lags, thus modifying amplitudes at all lags
to some extent. Unless the interference partial is very
strong, this is not sufficient to change the locations of the
Fig. 7. Error performance of AC2 PDA for various target instruments and pitch ranges with percussion instrument in background. (a)
Kick drum. (b) Hi hat. (c) Low agogo.
J. Audio Eng. Soc., Vol. 52, No. 4, April 2004
DETECTION OF MELODIC PITCH IN A PERCUSSIVE BACKGROUND
prominent peaks (at pitch and pitch multiples) but affects
only their relative amplitudes. As a result, the “choose the
highest peak in the ACF” approach typically results in
either a fine error due to a misshapen pitch peak or a gross
error in the form of a pitch octave error. Fig. 10(a) shows
the ACF of a periodic signal of fundamental frequency
600 Hz with the first four harmonics of amplitudes 10, 18,
14, and 12. At the sampling rate of 44.1 kHz, the signal
pitch period is 73.5 samples. A single interference partial
of fundamental frequency 3300 Hz and amplitude 16 (cor-
responding to signal-to-interference power ratio [SIR] 3.0)
is added to the signal, resulting in the ACF of the noisy
signal shown in Fig. 10(b). We see that the likelihood of
an octave error in the ACF of the noisy signal is highest
when, as depicted in Fig. 10, a valley of the ACF of the
385
RAO AND SHANDILYA
noise partial coincides with the signal pitch peak and the
peak of the noise ACF coincides with a target pitch mul-
tiple. This is true whenever an interference partial occurs
at or near an odd multiple of half the target fundamental
frequency. Likewise, we expect no pitch errors when the
interference partial is near a multiple of the target funda-
mental frequency. It is easy to see from Fig. 10 that the
likelihood of pitch octave error would increase as the am-
plitude of the noise partial increases relative to the target
signal strength. This explains why the introduction of a
linear filter such as the OEM filter affecting the relative
amplitudes of the signal and noise partials leads to a
change in the error profiles, as seen in Fig. 7. The intro-
duction of outer ear–middle ear filtering reduces the errors
in the case of the kick drum, but has the contrary effect in
Fig. 8. Error performance of AC3 PDA for various target instruments and pitch ranges with percussion instrument in background. (a)
Kick drum. (b) Hi hat. (c) Low agogo.
386
ENGINEERING REPORTS
the case of the hi hat and, to some extent, on the low
agogo. This can be explained by the low-frequency nature
of the spectrum of the kick drum, which consequently is
heavily attenuated by the OEM filter. The hi hat and low
agogo on the other hand have brighter spectra with much
middle frequency content that remains after the OEM fil-
ter. The target spectrum, because of its preponderant lower
harmonics, suffers greater overall attenuation than the bright
spectra percussions. The unusually sharp rise in pitch errors
in the high-pitch target range with hi-hat interference was
found to be due to the chance occurrence of an interference
partial at an odd multiple of half the fundamental frequency
of a note of recurring pitch throughout the song. This partial
fell near the resonance frequency (3 kHz) of the OEM filter
and was a prominent spectral component in the filtered sig-
J. Audio Eng. Soc., Vol. 52, No. 4, April 2004
ENGINEERING REPORTS
nal. It resulted in an octave error in the ACF pitch estimate
almost throughout the duration of the note in question.
Introducing the hair-cell model prior to ACF computa-
tion is equivalent to a nonlinear processing of the signal
that, among other effects, gives rise to new frequency
components located at sum and difference frequencies of
the original components. In the case of a weak or missing
fundamental, the creation of distortion components con-
tributes to the enhancement of the fundamental frequency
component [1]. In addition the hair-cell model introduces
a dc bias and a low-pass frequency selectivity [9]. The
presence of interference partials at nonharmonic locations
gives rise to nonharmonic distortion components, whose
magnitudes depend on the magnitudes and phases of the
interacting components (both signal and interference). Due
to this the distortion components affect the peak at pitch
lag in the ACF in different ways. One consistent effect is
the dc level introduced by the hair-cell processing that
Fig. 9. Error performance of the MH1 PDA for various target instruments and pitch ranges with percussion instrument in background.
(a) Kick drum. (b) Hi hat. (c) Low agogo.
J. Audio Eng. Soc., Vol. 52, No. 4, April 2004
DETECTION OF MELODIC PITCH IN A PERCUSSIVE BACKGROUND
leads to a bias favoring lower pitch lags, as seen in Fig.
10(c), where we also observe the attenuation of the high-
frequency partial. Such effects contribute to the overall
improvement in the performance demonstrated by the
AC3 PDA in Fig. 8. In particular, a more robust pitch
estimator is obtained in the case of the interference partial
at a high odd multiple of half the target fundamental fre-
quency. The Meddis–Hewitt PDA is an enhancement of
the AC3 algorithm in that a cochlear filter bank is in-
cluded. The ACF is computed separately in each fre-
quency channel, and summed across channels to obtain the
pitch estimate as the largest peak lag in the search range.
The frequency decomposition affected by the filter bank
limits the number of interacting partials through the hair-
cell model nonlinearity applied separately to each channel.
Fig. 11, obtained for the same signal and interference as
Fig. 10, illustrates the effect of this on the SACF. Shown,
for two different channel configurations (of four channels
387
RAO AND SHANDILYA
simulated by ideal bandpass filters), are the signal and
interference frequency components at the output of the
channel nonlinearities as well as the corresponding SACF.
In Fig. 11(a) the interference partial is in a separate chan-
nel by itself. This eliminates any distortion components
created due to an interaction of target harmonics and in-
terference. On the other hand, the co-occurrence of several
higher harmonics of the target in a single channel strength-
ens the fundamental frequency component in the SACF.
These two effects lead to an improved ACF peak at the
signal pitch period of 73 samples. This explains the im-
proved performance of the Meddis–Hewitt PDA for the
low agogo samples for the low- and middle-pitch ranges.
In the high-pitch range, however, it was observed that due
to the higher interharmonic spacing, several of the chan-
nels contained only a single harmonic of the target instru-
ment accompanied by interference components. This con-
dition is depicted and simulated by the configuration of
Fig. 11(b), where the signal partials occupy different chan-
nels and the noise partial shares a channel with a target
harmonic. The last channel gives rise to inharmonic dis-
tortion components, one of which is visible in the figure.
Together with the reduced contribution to the fundamental
frequency due to the absence of unresolved harmonics,
this leads to a degradation of the pitch estimate.
Finally we return to the AC1 PDA and explain the low-
frequency errors due to the kick drum in Fig. 6. The au-
tocorrelation method of AC1 leads to a significant number
Fig. 10. ACF plotted as a function of lag. (a) Signal of fundamental frequency 600 Hz (——) and noise tone of frequency 3300 Hz
(- - -). (b) Noisy signal. (c) Nonlinearly processed noisy signal.
388
ENGINEERING REPORTS
of gross errors for all three percussions, depending on the
frequency relation between target and interference partials.
The low-pitch-range errors are the most pronounced in the
case of the kick drum due to a strong low-frequency partial
from this percussion. Fig. 12 illustrates this effect on ACF
peak-based pitch estimation by simulating the kick drum by
a strong interference tone at 68 Hz. Shown in Fig. 12 are
ACFs for signals of fundamental frequency 600 Hz and 200
Hz with the same harmonic amplitudes as the signal of Fig.
10. Both clean signals yield accurate pitch peaks in the SACF
(at lags of 73 samples and 220 samples, respectively). How-
ever, the addition of the 68-Hz low-frequency interference
tone (with amplitude 26, corresponding to SIR 1.1) intro-
duces a low-lag bias in the overall ACF in both cases. This
leads to a gross pitch error (pitch submultiple selected) in
the case of the lower fundamental frequency signal since
its pitch period is comparable to that of the interference.
5 CONCLUSIONS
In this engineering report an experimental investigation
is presented of the performance of pitch-detection algo-
rithms based on temporal autocorrelation for the pitch
tracking of a melodic signal with percussive accompani-
ment characterized by inharmonic partials. The perfor-
mance of the autocorrelation pitch detector as well as its
enhancements based on the Meddis–Hewitt auditory
model are studied experimentally on synthetic musical sig-
nals. The ACF peak-based pitch detector incurs pitch es-
J. Audio Eng. Soc., Vol. 52, No. 4, April 2004
ENGINEERING REPORTS DETECTION OF MELODIC PITCH IN A PERCUSSIVE BACKGROUND

Fig. 11. SACF and spectral components of noisy signal after channel filtering and nonlinear processing corresponding to different
four-channel groupings of signal harmonics and noise. (a) SACF for /h1/h2/h3+h4/n/ and corresponding power spectrum. (b) SACF
for /h1/h2/h3/h4+n/ and corresponding power spectrum.

Fig. 12. ACF plotted as a function of lag for signal and interference tone of 68 Hz. - ⭈ - ⭈ - ACF of signal; - - - ACF of interference;
—— ACF of noisy signal. (a) Signal fundamental frequency 600 Hz. (b) Signal fundamental frequency 200 Hz.

J. Audio Eng. Soc., Vol. 52, No. 4, April 2004 389

RAO AND SHANDILYA
timation errors when interference partials cooccur with
signal harmonics. The following enhancements to the ba-
sic ACF PDA improve the robustness of pitch extraction
in the presence of percussive interference in the form of
inharmonic tones. Outer-ear-middle-ear prefiltering, co-
chlear bandpass filtering, and hair-cell nonlinear process-
ing represent a combination of linear and nonlinear pre-
processing of the signal before computing the ACF to
estimate pitch periodicity. The sum of channel autocorre-
lations (SACF) would simply be proportional to the auto-
correlation of the input signal to the cochlear filter bank if
it was not for the hair-cell nonlinearity [12]. Of signifi-
cance then is the combined role of channel separation and
hair-cell nonlinearity. The experimental results provide
important insights into the nature of the pitch errors and
their dependence on the relative frequencies of the signal
and interference. The noise sensitivity of ACF peak-based
pitch detection is highest when interfering partials fall ex-
actly between signal harmonics. The hair-cell nonlinearity
serves to increase the accuracy of pitch detection primarily
via an increased bias of lower lag peaks in the ACF and
the attenuation of high-frequency partials. The separate
processing of frequency channels by the introduction of
the cochlear filter bank is crucial in reducing the distortion
components from the interaction of signal harmonics and
interference partials while reinforcing the contribution to
the fundamental frequency component from unresolved
higher harmonics of the signal.
It would be interesting to explore the perceptual impli-
cations of the preceding observations. That is, since the
observations on the accuracy of the Meddis–Hewitt PDA
in the current context are explained by the signal-
processing algorithms used, it is relevant to wonder wheth-
er these specific predictions of the model hold for subjec-
tive pitch perception. From a practical viewpoint, the
results of this study may be applied to construct pitch
detectors for musical signals that are robust to the presence
of nonpitched percussion. It is interesting to consider tun-
ing the several available parameters of the hair-cell model
nonlinearity to increase its effectiveness for known signal
and interference characteristics, including possibly the
presence of broad-band noise. Alternative means of ob-
taining the SACF, such as summing with channel weight-
ing, provide further promising directions for future work.
THE AUTHORS
P. Rao
390
ENGINEERING REPORTS
6 REFERENCES
[1] W. Hess, Pitch Determination of Speech Signals
(Springer, New York, 1983).
[2] J. D. Wise, J. R. Capiro, and T. W. Parks, “Max-
imum Likelihood Pitch Estimation,” IEEE Trans.
Acoustics, Speech, Signal Process., vol. ASSP-24 (1976
Oct.).
[3] G. Monti and M. Sandler, “Monophonic Transcrip-
tion with Autocorrelation,” in Proc. COST G-6 Conf. on
Digital Audio Effects, DAFX-00 (Verona, Italy, 2000).
[4] D. J. Hermes, “Pitch Analysis,” in Visual Represen-
tations of Speech Signals, M. Cooke, S. Beet, and M.
Crawford, Eds. (Wiley, New York, 1993).
[5] R. Meddis, and M. J. Hewitt, “Virtual Pitch and
Phase Sensitivity of a Computer Model of the Auditory
Periphery. I: Pitch Identification,” J. Acoust. Soc. Am., vol.
89 (1991 June).
[6] M. Pflueger, R. Höldrich, and W. Riedler, “A Non-
linear Model of the Peripheral Auditory System,” pre-
sented at the 103rd Convention of the Audio Engineering
Society, J. Audio Eng. Soc. (Abstracts), vol. 45, p. 1002
(1997 Nov.), preprint 4510.
[7] “HUTear—Matlab Toolbox for Auditory Model-
ing,” available at www.acoustics.hut.fi/software/HUTear/.
[8] A. Klapuri, “Wide-Band Pitch Estimation for Natu-
ral Sound Sources with Inharmonicities,” presented at the
106th Convention of the Audio Engineering Society, J.
Audio Eng. Soc. (Abstracts), vol. 47, p. 520 (1999 June),
preprint 4906.
[9] R. Meddis, “Simulation of Mechanical to Neural
Transduction in the Auditory Receptor,” J. Acoust. Soc.
Am., vol. 79 (1986, Mar.).
[10] R. Meddis, M. Hewitt, and T. Shackleton, “Imple-
mentation Details of a Computation Model of the Inner
Hair-Cell/Auditory-Nerve Synapse,” J. Acoust. Soc. Am.,
vol. 87 (1990 Apr.).
[11] M. Slaney, Auditory Toolbox, version 2, Interval
Research Corporation.
[12] P. Cariani, M. Tramo, and B. Delgutte, “Neural
Representation of Pitch through Temporal Autocorrela-
tion,” presented at the 103rd Convention of the Audio
Engineering Society, J. Audio Eng. Soc. (Abstracts), vol.
45, pp. 1021–1022 (1997 Nov.), preprint 4583.
S. Shandilya
J. Audio Eng. Soc., Vol. 52, No. 4, April 2004
ENGINEERING REPORTS
Preeti Rao received a Bachelor degree in Electrical
Engineering from the Indian Institute of Technology,
Bombay, in 1984, and a Ph.D. degree specializing in
signal processing from the University of Florida, Gaines-
ville, in 1990. She taught in the Electrical Engineering
department at the Indian Institute of Technology, Kanpur
from 1994 to 1999. Following a six-month visiting posi-
tion at the Institute of Perception Research, Eindhoven,
The Netherlands, she joined the Indian Institute of
Technology, Bombay, where she is presently an associ-
ate professor. Her current research interests include
J. Audio Eng. Soc., Vol. 52, No. 4, April 2004
DETECTION OF MELODIC PITCH IN A PERCUSSIVE BACKGROUND
speech and audio signal compression and audio content
retrieval.
●
Saurabh Shandilya received a B.E. degree in electrical
engineering from the Government Engineering College, Bi-
laspur, India, in 2001 and an M.Tech. degree in electrical
engineering from the Indian Institute of Technology, Bom-
bay in 2003. Currently he works for Neomagic Semiconduc-
tors Inc., Noida, India. His interests include speech and audio
processing, video coding, and associative computing.
391
 AES STANDARDS
COMMITTEE NEWS
Detailed information regarding AES Standards Committee
(AESSC) proceedings including structure, procedures, reports,
meetings, and membership is published on the AES Standards Web
site at http://www.aes.org/standards/. Membership of AESSC work-
ing groups is open to any individual materially and directly affect-
ed by the work of the group. For current project schedules, see the
project-status document also on the Web site.

climate following AESSC has rightfully earned a high

New AESSC Chair Named
I am delighted to announce that Richard
Chalmers has been appointed as the
new AES Standards Committee Chair
by the AES President, Ron Streicher.
You will recall that, just before the
AES 115th Convention in New York,
John Nunn notified us, “[T]hat for
personal reasons I am unable to
consider re-appointment when my
current term as AESSC Chair comes to
an end.” After extensive research and
consideration, this announcement
provides the AESSC with a clear path
into the future.
Richard Chalmers writes:
“I am pleased and honored to take
over from John Nunn as chair of the
AESSC. For many years I was secretary
to a number of EBU audio technical
groups and had the pleasure of meeting
several AESSC chairs as well as
keeping in touch with AES work on a
more formal basis through John and
Mark Yonge.
“Of course the interests of AES
members are wider (and deeper!) than
those of the EBU broadcasters so I am
expecting to follow a learning curve on
a number of subjects.
“I am very conscious that the AESSC
has to operate in a more restrictive
For its published documents and
reports the AESSC is guided by Inter-
national Electrotechnical Commission
(IEC) style as described in the ISO/IEC
Directives, Part 3. IEC style differs in
some respects from the style of the AES
as used elsewhere in this Journal.
AESSC document stages referenced
are: Project initiation request (PIR);
Proposed task-group draft (PTD);
Proposed working-group draft (PWD);
Proposed call for comment (PCFC);
Call for comment (CFC)
the problems
resulting from
September 11,
2001, but I hope
that the standards
work of the AES
Richard Chalmers
will continue to
reflect the broad interests of the
members and be driven by their needs.
“I look forward to meeting everyone
and helping as much as I can to forward
the work of the AES Standards
Committee.
“Richard Chalmers”
John Nunn, outgoing chair, writes:
“Just before the New York con-
vention I informed the AESSC that I
would be standing down as its Chair
once the appointment of a new Chair
could be made by the AES President,
Ron Streicher. I have known Richard for
many years and I am confident that his
knowledge of audio and his experience
in the standards business will ensure the
future of AES Standards in safe hands.
“When I first became involved with
the AES standards activities, more than
twenty years ago, I never considered
that one day I might become Chair of
the AESSC and it has been a privilege
to serve the Society in this capacity for
some nine years. Yes, it has been hard
work at times but it has also been very
enjoyable and satisfying leading the
AESSC through a period of consid-
erable evolution. The rapid techno-
logical changes in the audio industry
have produced many challenges, both
technical and structural, for the AESSC
and I should like to thank the AES
Board of Governors for their support
during this period. I should also like to
thank all the members of the AESSC
who have worked on the standards
which have been published and who
have given me their support. The
reputation in the international standards
world and I believe it is well placed to
continue to be respected for years to
come.
“I wish the AESSC well and I hope
Richard Chalmers has as much fun I had.
“John Nunn”
A note from Mark Yonge, AES
Standards Secretary:
I would like to express here the
appreciation of everybody involved in
the Standards Committee for John’s
stewardship and astonishing hard work
through some very interesting times. In
the past nine years, the Standards
Committee has moved from a paper-
based operation to a modern, internet-
centered organization; the AES has
acquired an increased international sig-
nificance through its relationship with
the IEC; on-line communications now
permit faster document development
than at any time in the past.
From a personal point of view, I owe
a huge debt of gratitude to John for the
support he gave to both Dan Queen and
myself during the nontrivial process of
secretariat transition in 2002.
Richard Chalmers’s own activities
within the European Broadcasting
Union (EBU) over a number of years
shows him to be someone who under-
stands the purpose and value of audio
standards as well as someone who
brings an abundance of knowledge and
experience to the work of the AESSC.
Please join me in welcoming Richard
to the AES Standards Committee. I very
much look forward to working with him
in the times that lie ahead.
Newly Published
AES-R2, AES project report for
articles on professional audio and
for equipment specifications—

392 J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April

 AES STANDARD
COMMITTEE NEWS
Notations for expressing levels has
been revised.
The revision has clarified the use of
the “dBu” and abandons the earlier
intention to recommend the adoption of
1 V as the reference quantity for audio
levels in decibels.
The revised document can be found
on the Web site.
AES14-1992 AES standard for
professional audio equipment—
Application of connectors, Part 1:
XLR-type polarity and gender has been
reaffirmed.
AES17-1998 AES standard method
for digital audio equipment—
Measurement of digital audio
equipment has been reaffirmed.
Call for Comment
The following document will be
withdrawn by the AES after any adverse
comment received within three months
of the publication of their call on the
AES Standards Web site has been
resolved. For more detailed information
on each call please go to
www.aes.org/standards/b_comments.
Comments should be sent by
e-mail to the secretariat at
standards@aes.org. All comments will
be published on the Web site.
Persons unable to obtain this
document from the Web site may
request a copy from the secretariat.
CALL FOR COMMENT on with-
drawal of AES33-1999, AES standard
procedure for maintenance of AES
audio connector database has been
published, 2004-02-13.
New Project AES-X144 on
DSD over AES47
Direct stream digital (DSD)
audio coding is becoming estab-
lished within the industry as an
alternative to linear PCM
c o d i n g f o r p r o f e s sional audio
applications. Project AES-X140 is
intended to standardize a transport
which will carry both linear PCM and
DSD. If AES47 is not extended to
include DSD in a standardized way, a
variety of incompatible implemen-
tations can be anticipated.
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April
Project AES-X144 Carriage of DS
Audio Data in AES47
Scope: “[T]o amend AES47-2002
clauses 4.1.2.1, 5.2.2.2 and 6 to provide
the option of transmitting DSD audio
instead of linear PCM.”
Summary Report: SC-06-01
Working Group on Audio-
File Transfer and Exchange
This meeting was held in conjunction
with the AES 115th Convention, NY,
2003-10-10 and was convened by chair
M. Yonge.
DVD Forum Liaison
J. Yoshio reported from DVD Forum
Working Group 10. This group is con-
sidering professional applications of
DVD optical disc.
There is currently no standard for pro-
fessional audio recording on DVD.
Application areas include studios and
broadcast. The intention is to consider a
new format for recording and playback
of high quality audio data and associated
data. The next generation of DVD tech-
nology will handle the higher bandwidth
and storage capacity necessary for: a)
timecode; b) metadata; c) video
recording possibilities; d) audio
recording for professional use.
Two possibilities are being con-
sidered for development as the DVD
Audio Professional (DVD-AP) format:
a) Broadcast Wave Format (BWF), a
computer file format on DVD-ROM; b)
a DVD audio format dedicated for pro-
fessional audio application.
It could be possible to have two
sessions on a single DVD disc: DVD
audio plus a DVD-ROM area. However,
this will need a special DVD AP
recorder/player.
It had been noted that BWF files will
be a standard for broadcasters. The
Japanese Post Production Association
(JPPA) had promoted the use of the
BWF-J format on 2.5 in MO disc, which
was very popular in Japan
although not so widely used elsewhere.
However, the capacity of an MO disc
was not enough for multichannel audio
files. Further development of this style
of interchange would require the
capacity of the DVD-AP instead of the
MO disc.
AES31-3 Audio file transfer and
exchange—Part 3: Simple project
interchange
Mix automation
A proposed extension of AES31-3 is
intended to provide a simple but
reliable method of interchanging gain
and pan automation data between
workstations.
Should there be some maximum gain
in a fader, referenced to unity? U. Henry
observed that many workstations have
difficulty applying more than 12 dB
gain to an audio path; sometimes less. It
would be helpful if it was generally
understood that this was a realistic con-
straint on free interchange. Bull agreed
but noted that replay systems should be
prepared to accommodate larger gain
maxima in case these are encountered in
practice.
There was a discussion concerning
gains in decibels expressed to two
decimal places. Henry noted that
because AES31-3 requires the value to
be rendered in ASCII characters, some
limit to the number of characters is prac-
tically necessary. It was noted also that
receiving applications will need to inter-
polate, or ramp, between gain steps in
any case, and this will largely eliminate
the need for higher precision gain
values; two decimal places were felt to
be sufficient.
Panning
Bull felt that 100 points in each
panning axis (left to right and front to
back) would not be sufficient to define
a smooth pan locus. Imagine a circular
pan movement; there could be audible
path errors, or gain artifacts similar to
zipper noise. With such coarse steps,
eliminating these artifacts would need a
greater degree of look-ahead to up-
coming data which could make
practical implementations unneces-
sarily complex.
It was felt that the default pan
position should be at front-center,
which should be identified by zero
coordinates in both axes.
It was also felt that there needed to be
a convention for interpreting pan points
in a similar manner to that proposed for
faders. This should avoid the ap-
pearance of instantaneous tran- ➥
393
 AES STANDARD
COMMITTEE NEWS

sitions—implementations should always
use ramped coefficients in the same way
as the faders.
Following a question about panning
law, there was agreement that the
document should define the total SPL in
the room from all loudspeakers to be
constant, independent of the pan
position.
New business
S. Aoki spoke on the subject of
Broadcast Wave Format (BWF) files in
international interchange.
BWF as currently defined does not
support international interchange
because, for example, the metadata in
the “bext” chunk is specied to use the
ASCII character code. While ASCII is a
robust and simple character code, it is
limited to representing the Roman
alphabet. ASCII code is not useful in
Japanese operations because operators
cannot read information in their own
language. Interchange becomes difficult
when interchanging commercials for
broadcasting where flexibility of
interchange is very important.
Aoki felt that Japanese character sets
based on ISO 10646 Universal Multi-
Octet Coded Character Set (UCS) or its
derivative UCS Transformation Format,
UTF-16, would be appropriate. Unicode
and the Japanese-specific “Shift-JIS” are
also in use. In the Japan industry there
appeared no clear consensus as to which
character code to prefer.
It was pointed out that any general
standard should probably cover all
international interchange and support all
character sets, not just Japanese. UTF-
16 appears to be derived from ISO
10646; Unicode may be considered a
subset of this same international
standard.
Bull noted that it would be a sig-
nificant task to convert existing appli-
cations to a different character code set.
Also, character codes sets are not inter-
operable so there will need to be some
degree of active translation.
Henry observed that multi-octet file
names will be a problem; they may not
be readable at all in some systems! Aoki
said that the Japan Post Production
Association (JPPA) has no special re-
quirement but uses the file name scheme
provided in the Japanese version of the
computer operating system.
Henry remained concerned that file
names will not be read correctly—
AES31-3 depends on locating files by
file names. Multibyte file names could
corrupt accurate file name reading.
This issue has implications that will
need further consideration.
Summary Report: SC-06-02
Working Group on Audio
Applications Using the High
Performance Serial Bus
(IEEE 1394)
This meeting was held in conjunction
with the AES 115th Convention, New
York, 2003-10-10 and was convened
by chair J. Strawn.
Liaison with 1394 Trade Association
(1394 TA)
Fujimori reported as follows:
The A/V Working Group is looking
at the following topics:
• Blu-Ray DVD recording.
• Japanese terrestrial digital television
(DTV).
• A point-to-point test network.
• AV/C Camera storage subunit 2.1.
• IEEE Ballot Review Committee
(BRC) review of 1394 implementation
tests.
• S800 Base-T, now known as IEEE-
1394c.
• Pin assignments on UTP5 cable.
The proposal from Digital Home
Technology is to use pins 4 and 5 for
48V, and 3 and 6 for ground. B.
Moses will get more information on
this.
M. Mora of Apple asked if the
Trade Association plans to publish
audio guidelines similar to video
guidelines. Fujimori answered that the
DVD Forum and other organizations
are expected to do that within their
application space.
New projects
Strawn will submit a project initiation
form for General Data based on IEC
61883-6.

Proceedings of the
AES 24th International Conference:
THE PROCEEDINGS Multichannel Audio, The New Reality
OF THE AES 24 th
INTERNATIONAL Banff, Alberta, Canada
CONFERENCE 2003 June 26-28.
This conference was a follow-up to the 19th Conference on surround sound.
2003 June 26–28 These papers describe multichannel sound from production and engineering
to research and development, manufacturing, and marketing. 350 pages

Also available on CD-ROM

You can purchase the book and CD-ROM online at

www.aes.org.
Banff, Alberta, Canada For more information email Andy Veloz at
AAV@aes.org
Conference Chair:
Theresa Leonard
or telephone +1 212 661 8528 x39.

394 J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April

 REVIEWS OF
ACOUSTICAL PATENTS*
Any opinions expressed here are those of reviewers as
individuals and are not legal opinions. Printed copies
of United States Patents here reviewed may be ordered
at $3.00 each from the Commissioner of Patents and
Trademarks, Washington, D.C. 20231. Patents are
available via the Internet at http://www.uspto.gov.

LLOYD RICE
11222 Flatiron Drive, Lafayette, Colorado 80026
REVIEWERS
GEORGE L. AUGSPURGER, Perception Incorporated, MARK KAHRS, Department of Electrical Engineering,
Box 39536, Los Angeles, California 9003 University of Pittsburgh, Pittsburgh, Pennsylvannia 15261

6,535,610 in ma gnetic gap 57 formed be tween ring magnet 11 and pole piece 52. It
seems obvious that the magnet must have one pole on its inner surface
43.38.Hz DIRECTIONAL MICROPHONE UTILIZING facing the voice coil and the other pole on its outer surface abutting the pole
SPACED APART OMNI-DIRECTIONAL
MICROPHONES
Brett B. Stewart, assignor to Morgan Stanley & Company
Incorporated
18 March 2003 „Class 381Õ92…; filed 7 February 1996
The c urrent buzzword in audio pickup for teleconferencing is ‘‘beam-
forming.’’ If mere beamforming is not up to the task, then ‘‘adaptive beam-
forming’’ will surely do the trick. This latest invention mounts a few omni-
directional microphones around the periphery of a video display, digitizes
their outputs, and then processes the signals through tapped delay lines
under computer control.—GLA
piece. According to thispatent, a more linear magnetic field can be achieved
if ma gnetization is divided into three areas. Only the central portion of the
6,542,614 ring is magnetized at right angles. The upper and lower portions are mag-
netized at angles of 40° or less.—GLA
43.38.Si BOOMLESS HEARINGÕSPEAKING
CONFIGURATION FOR SOUND RECEIVING MEANS
6,542,617
Heinz Renner, assignor to Koninklijke Philips Electronics N.V.
1 April 2003 „Class 381Õ370…; filed in the European Patent Office
43.38.Ja SPEAKER
21 March 2001
Certain hands-free communication applications require the user to hear Masao Fujihira et al., assignors to Sony Corporation
local sounds as well as incoming signals. This requirement can be met by a 1 April 2003 „Class 381Õ402…; filed in Japan 26 May 1999
single headphone having an attached boom microphone. Although the mi-
What appears to be a conventional self-shielded l oudspeaker is in fact
crophone is advantageously close to the user’s mouth, its proximity results
in unwanted pickup of pops and air noises as well. Moreover, the assembly a s mall unit designed to reproduce frequencies up to 70 kHz or so. Voice
is easily dislodged if the user is moving. This patent describes a lightweight, coil bobbin 11 is made of a conductive material such as aluminum. Coil 6 is
clip-on earpiece that contains not only a headphone but an embedded direc- attached to the bobbin by a soft bonding agent that decouples the coil from
tional microphone.—GLA

6,529,107
43.38.Ja SPEAKER COMPRISING RING MAGNET
Motoharu Shimizu and Hiroyuki Daichoh, assignors to Hitachi
Metals Limited
4 March 2003 „Class 335Õ302…; filed in Japan 16 December 1999
Modern magnetic materials allow moving coil loudspeakers to be built
with thin, lightweight ring magnets. In the illustration, voice coil 55 moves

* The editors of this Journal are grateful to the Journal of the

Acoustical Society of America for allowing us to republish, from
their monthly Reviews, selected patents of interest to our readers.
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April 395
 REVIEWS OF
ACOUSTICAL PATENTS*
the bobbin at very high frequencies. The patent explains that in this very
high range the bobbin is driven as an induction motor and continues to
radiate sound, presumably from dust cap 15.—GLA
6,556,687
43.38.Hz SUPER-DIRECTIONAL LOUDSPEAKER
USING ULTRASONIC WAVE
Koji Manabe, assignor to NEC Corporation
29 April 2003 „Class 381Õ387…; filed in Japan 23 February 1998
It is known that an array of ultrasonic transducers can be driven by a
modulated carrier to produce audible sound from empty space. This patent
suggests that if the transducers are mounted on a concave surface, then their
combined energy can be focused at a specific point in space. Moreover, if
the curvature is adjustable, then the focal point can be shifted. These two
observations would seem to be self-evident, but they were expanded into 17
patent claims.—GLA
6,535,269
43.38.Md VIDEO KARAOKE SYSTEM AND
METHOD OF USE
Gary Sherman and M ichael Ch ase, both of Lo s Angel es,
California
18 M arch 2003 „Class 352 Õ6…; filed 29 June 2001
The sound track of a commercial motion picture is already created and
stored in a multi-track format. Individual tracks may be rerecorded later to
correct audio problems or to dub dialog into another language. Suppose that
you could purchase a DVD for home viewing that allowed you to record and
replace selected dialog tracks with your own overdubs. The patent describes
an interactive system to facilitate such customized viewing.—GLA
396
6,554,098
43.38.Ja PANEL SPEAKER WITH WIDE FREE
SPACE
Tatsumi Komura, assignor to NEC Corporation
29 April 2003 „Class 181Õ173…; filed in Japan 15 June 1999
To save space, a panel-type loudspeaker diaphragm 1 can be driven at,
or very near, its outer edge. A single panel can be driven by more than one
transducer 2, 2⬘.—GLA
6,543,574
43.38.Ja METHOD OF MAKING A SPEAKER EDGE
CONTAINING ISOCYANATE AND POLYOL
Sinya Mizone et al., assignors to Inoac Corporation; Matsushita
Electric Industrial Company, Limited
8 April 2003 „Class 181Õ171…; filed in Japan 9 March 1999
A half-roll loudspeaker cone suspension is usually formed from sheet
stock molded to shape by heat and pressure. The patent describes an inter-
esting alternative in which a liquid material is injected into a die cavity and
then subjected to reaction foaming and curing—a little like baking a
waffle.—GLA
6,549,632
43.38.Kb MICROPHONE
Hiroshi Akino et al., assignors to Kabushiki Kaisha
Audio-Technica
15 April 2003 „Class 381Õ174…; filed 19 March 1997
Some hand-held microphones are extremely sensitive to mechanical
hocks and scrapes. This patent describes a simple, passive shock isolation
ystem derived from mechanical analog circuit analysis. Although the patent
ext describes embodiments for both dynamic and capacitor microphones,
all of the eight patent claims refer to capacitor microphones only.—GLA
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April
 REVIEWS OF
ACOUSTICAL PATENTS*

6,535,610
43.38.Hz DIRECTIONAL MICROPHONE UTILIZING
SPACED APART OMNI-DIRECTIONAL
MICROPHONES
Brett B. Stewart, assignor to Morgan Stanley & Company
Incorporated
18 March 2003 „Class 381Õ92…; filed 7 February 1996
The c urrent buzzword in audio pickup for teleconferencing is ‘‘beam-
forming.’’ If mere beamforming is not up to the task, then ‘‘adaptive beam-
forming’’ will surely do the trick. This latest invention mounts a few omni- quite different from a standard closed-box loudspeaker. Some general rules
directional microphones around the periphery of a video display, digitizes for predicting and optimizing the performance of such a system are devel-
their outputs, and then processes the signals through tapped delay lines oped and explained.—GLA
under computer control.—GLA

6,549,629
43.38.Tj DVE SYSTEM WITH NORMALIZED
6,549,637
SELECTION
43.38.Ja LOUDSPEAKER WITH DIFFERENTIAL
Brian M. Finn and Shawn K. Steenhagen, assignors to Digisonix
FLOW VENT MEANS LLC
15 April 2003 „Class 381Õ92…; filed 21 February 2001
Jon M. Risch, assignor to Peavey Electronics Corporation
15 April 2003 „Class 381Õ397…; filed 24 September 1998 DVE stands for digital voice enhancement which, in this case, includes
echo cancellation, background noise suppression, and optimal selection of
This patent includes just a little bit of everything, culminating in 40
multiple-zone microphones in a hands-free communications system.
fairly lengthy claims. However, the heart of the invention is the differential
flow vent shown. This is an open-ended cylinder 100 fitted with funnel-
shaped insert 104. We are informed that air flowing from left to right will

Although the procedure is too complicated to describe in a few sentences, it

encounter more resistance than that flowing in the reverse direction. If two
conventional vents in a woofer box are replaced by differential vents of
opposite polarity, the result is forced air ventilation—benign turbulence, so
to speak.—GLA
6,553,124
43.38.Ja ACOUSTIC DEVICE
Henry Azima and Joerg Panzer, assignors to New Transducers
Limited
22 April 2003 „Class 381Õ345…; filed in the United Kingdom 2 Sep-
tember 1995
This is an interesting patent that includes more than 20 pages of actual
test results. The document is really a research paper followed by seven
patent claims. The patent teaches that enclosing the rear radiation of a bend-
ing wave, panel-type loudspeaker results in a system that is acoustically
is clearly explained in the patent. Anyone interested in the field will find the
patent informative.—GLA
6,570,079
43.38.Md RECORDING AND REPRODUCING
APPARATUS, RECORDING AND REPRODUCING
METHOD, AND DATA PROCESSING
APPARATUS
Shinichi Fukuda, assignor to Sony Corporation
27 May 2003 „Class 84Õ602…; filed in Japan 19 February 1998
Sony let the examiner do the legwork on this one. Suppose an audio
CD owner wishes to make a copy. If a lossy copy is chosen, then the user is
stuck with just a bad fidelity copy but not a bill. If the high-fidelity copy is
chosen, then the ‘‘accounting system’’ is involved and the owner is charged
a fee. The patent doesn’t address any of the wider issues regarding music
distribution.—MK

J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April 397

 REVIEWS OF
ACOUSTICAL PATENTS*

6,557,665
43.38.Ja ACTIVE DIPOLE INLET USING DRONE
CONE SPEAKER DRIVER
Richard D. McWilliam and Ian R. McLean, assignors to Siemens
Canada Limited
6 May 2003 „Class 181Õ206…; filed 16 May 2001
The invention is intended to provide active noise cancellation at the air
intake of an internal combustion engine. Inner diaphragm 18 is electrically
driven by conventional means 46. Outer diaphragm 22 is driven acoustically
to generate a noise attenuating signal. At the same time, air is somehow

flowing from mouth 38 in direction A through passageway 14 into the

engine even though both diaphragms are provided with seals 30, 34. The
patent claims do not clarify the arrangement.—GLA

6,535,613
43.38.Ja AIR FLOW CONTROL DEVICE FOR
LOUDSPEAKER
Jason A. Ssutu, assignor to JL Audio, Incorporated
18 March 2003 „Class 381Õ397…; filed 28 December 1999
Airtight dust cap 44 pumps air in and out of cavity
generated in a sound wave amplifying horn to increase amplification effi-
ciency of bass sounds and improve the clearness of the resulting sounds.’’—
GLA
46 through rear 6,557,664
43.38.Ja LOUDSPEAKER
Anthony John Andrews and John Newsham, both of Dorking,
Surrey, the United Kingdom
6 May 2003 „Class 181Õ152…; filed 22 February 1994
Central plug 21 is actually chisel-shaped, and surrounding horn 11 is
similarly asymmetrical. The overall assembly is a close cousin to the JBL
2405 high-frequency transducer designed more than 25 years ago. In both

vent 31. Plate 52 directs the air flow against the inner surface of bobbin 35
to c ool voice coil 36.—GLA

6,560,343
43.38.Ja SPEAKER SYSTEM
Jae-Nam Kim, assignor to Samsung Electronics Company,
Limited
6 May 2003 „Class 381Õ349…; filed in the Republic of Korea 22
April 1996
Part of the backwave energy from loudspeaker 16 is conducted
through horn 24 to the face of cabinet 10. The remainder energizes vent 26.
The patent explains that since only a portion of the rear sound waves are cases the objective is to create a coverage pattern that is relatively wide
collected and amplified, ‘‘...reflected waves or standing waves will not be horizontally but vertically narrow.—GLA

398 J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April

 REVIEWS OF
ACOUSTICAL PATENTS*

6,546,105 rectified, and then divided—not compared. The patent explains in some
detail how this arrangement estimates source proximity rather than signal-
43.38.Vk SOUND IMAGE LOCALIZATION DEVICE to-noise ratio. If the proximity estimation signal exceeds a predetermined
AND SOUND IMAGE LOCALIZATION threshold, then microphone 310 is gated on or its gain raised.—GLA
METHOD
6,553,121
Takashi Katayama et al., assignors to Matsushita Electric
Industrial Company, Limited
43.38.Vk THREE-DIMENSIONAL ACOUSTIC
8 April 2003 „Class 381Õ17…; filed in Japan 30 October 1998
PROCESSOR WHICH USES LINEAR PREDICTIVE
Using head-related transfer functions 共HRTFs兲 to create virtual sound
sources from a pair of loudspeakers is theoretically intriguing but messy in
COEFFICIENTS
practice. Assuming that some kind of all-purpose HRTFs can be derived,
Naoshi Matsuo and Kaori Suzuki, assignors to Fujitsu Limited
then FIR filter coefficients can be calculated for any angular location. How-
22 April 2003 „Class 381Õ17…; filed in Japan 8 September 1995
ever, computing and/or storing filter coefficients for all possible locations is
inefficient and time-consuming. This patent, like a number of earlier inven- To create convincing three-dimensional audio for computer games, a
number of virtual sound sources must be controlled within a virtual sound
field. At the same time, the acoustical characteristics of the actual reproduc

tions, attempts to find a better way. In this case, the angular location of a
virtual source is fed to a coefficient control device which then performs
digital mathematical operations involving only three predetermined fre-
quency response functions. The process is said to result in a dramatic reduc-
tion in memory requirements and computational time, as compared to prior
art.—GLA

6,549,630
43.38.Lc SIGNAL EXPANDER WITH
DISCRIMINATION BETWEEN CLOSE AND
DISTANT ACOUSTIC SOURCE
James F. Bobisuthi, assignor to Plantronics, Incorporated
15 April 2003 „Class 381Õ94.7…; filed 4 February 2000
The objective is to reliably turn on a microphone when its user speaks
and to minimize false triggering from other sound sources. A handset or
headset is fitted with two microphones, one 310 near the talker’s mouth and
the other 330 as far as possible from the first. Their outputs are filtered,
ing sound field must be subtracted from the sound source. A bank of large
FIR filters seems to be called for, but the patent suggests a more efficient
approach. By performing linear predictive analysis of the impulse response
of the sound field to be added, the number of taps can be greatly reduced. A
similar procedure can be applied to sound sources in motion—in effect,
panning between locations rather than creating a plurality of individual
sources. The patent is clearly written and includes a great many helpful
illustrations.—GLA
6,549,627
43.38.Lc GENERATING CALIBRATION SIGNALS
FOR AN ADAPTIVE BEAMFORMER
Jim Rasmusson et al., assignors to Telefonaktiebolaget LM
Ericsson
15 April 2003 „Class 381Õ71.11…; filed 30 January 1998
Consider a hands-free communications system installed in a vehicle.
The equipment includes two or more microphones 405, 407 and a loud-
speaker 401. By introducing adaptive filters at the outputs of individual
microphones it is possible to achieve in-phase summation of the signals
from the direction of a talker while largely canceling the unwanted signals
from the loudspeaker. Because the acoustical environment and the location

J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April 399

 REVIEWS OF
ACOUSTICAL PATENTS*

6,577,736
43.38.Vk METHOD OF SYNTHESIZING A THREE
DIMENSIONAL SOUND-FIELD
Richard David Clemow, assignor to Central Research
Laboratories Limited
10 June 2003 „Class 381Õ18…; filed in the United Kingdom 15 Oc-
tober 1998
A home surround sound system typically uses two or three front speak-
ers plus two rear speakers. When mixing program material for this format,
conventional panning techniques work well when moving sound images
laterally but cannot create accurate phantom sources between front and rear
speakers. Conversely, by making use of head-related transfer functions and

of the talker may change, the system must somehow calibrate itself. The
patent describes an improved one-step calibration process that may also be
augmented by utilizing the filters as fixed echo cancellers during normal
operation.—GLA

6,563,932
43.38.Ja MAGNET SYSTEM FOR LOUDSPEAKERS
Paul Cork, assignor to KH Technology
13 May 2003 „Class 381Õ412…; filed in the United Kingdom 16
January 2001
An electrodynamic loudspeaker magnetic circuit has a ring-shaped gap
42 between inner and outer pole pieces 26 and 36. Typically, the gap would

interaural crosstalk cancellation, a two-speaker playback system can theo-

retically locate sound images almost anywhere if it is designed for one
particular listener at a specific location. The inventor has developed an in-
teresting method for combining the best features of both systems. According
to the patent, it can easily be combined with conventional multi-channel
mixing techniques.—GLA

6,587,565
be energized by a single magnet 20. The addition of complementary magnet
44 is said to overcome deficiencies of prior art in terms of reduced size, 43.38.Vk SYSTEM FOR IMPROVING A SPATIAL
improved performance, and ease of assembly.—GLA EFFECT OF STEREO SOUND OR ENCODED
SOUND
Pyung Choi, assignor to 3S-Tech Company, Limited
1 July 2003 „Class 381Õ98…; filed in the Republic of Korea 13
6,574,339 March 1997

43.38.Vk THREE-DIMENSIONAL SOUND This patent describes yet another stereo enhancement method based on
filtering and cross-coupling. In this case, signal processing is applied to left
REPRODUCING APPARATUS FOR MULTIPLE
LISTENERS AND METHOD THEREOF
Doh-hyung Kim and Yang-seock Seo, assignors to Samsung
Electronics Company, Limited
3 June 2003 „Class 381Õ17…; filed 20 October 1998
Using head-related transfer functions and some fancy digital filtering,
it is possible to create a convincing three-dimensional sound field from two
loudspeakers. A major drawback is that the illusion is effectively limited to
a single listener at a defined location. We might give the listener a choice of
locations by allowing him to select from, say, three filter settings. But sup-
pose that all three settings are selected sequentially at some optimum time
interval. Will the result be aural confusion or will the effective sound field and right channels individually. The original left and right signals are ‘‘en-
be expanded to accommodate multiple listeners? The patent argues for the hanced’’ at low frequencies but otherwise unmodified. The difference signal
latter.—GLA is ‘‘enhanced’’ in the high-frequency range.—GLA

400 J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April

 2003
ROUGH
D TH
UPDATE 10,000
Journal technical
articles, convention
preprints, and
conference papers
at your fingertips

The Audio Engineering Society has published a 20-disk electronic library containing most of the
Journal technical articles, convention preprints, and conference papers published by the AES since
its inception through the year 2003. The approximately 10,000 papers and articles are stored in PDF
format, preserving the original documents to the highest fidelity possible while permitting full-text
and field searching. The library can be viewed on Windows, Mac, and UNIX platforms.
You can purchase the entire 20-disk library or disk 1 alone. Disk 1 contains the program and
installation files that are linked to the PDF collections on the other 19 disks. For reference and
citation convenience, disk 1 also contains a full index of all documents within the library, permit-
ting you to retrieve titles, author names, original publication name, publication date, page num-
bers, and abstract text without ever having to swap disks.

For price and ordering information send email to Andy Veloz at

aav@aes.org, visit the AES web site at www.aes.org, or call any
AES office at +1 212 661 8528 ext. 39 (USA);
+44 1628 663725 (UK); +33 1 4881 4632 (Europe).
 AES 25 INTERNATIONAL th
CONFERENCE
n the 135-meter-high London
Eye you can see as far as 25
miles away, and you have a
bird’s eye view of such major
London sights as St. Paul’s
Cathedral, Buckingham Palace,
the Houses of Parliament, and
Big Ben. Chair John Grant and
his committee are planning a con-
ference this June 17–19 that will
give you a great view of the critically
important subject of metadata. Metadata for
Audio will be held at Church House, the conference center that
is just a stone’s throw from Westminster Abbey and the
Houses of Parliament in central London.
As the means for production and distribution of digital audio
proliferate, appropriate metadata tools are needed to facilitate,
control, and extend these activities. There has been a great deal
of activity in individual organizations to develop metadata
tools. However, substantial issues remain to be addressed
before the desired goal of global exchange and common under-
standing can be reached. International standardization, such as
the work on MPEG-7 and MPEG-21 may hold some important
Photos courtesy British Tourist Office
answers.
This conference seeks to describe the state of the art, identify
the issues, and indicate directions for the develop-
ment of advanced metadata systems, both for con-
sumer distribution and business-to-business. It will
bring together media publishers and software
designers, media librarians and archivists, database
managers and streaming engineers whose opera-
tions are increasingly dependent on the success of
sophisticated metadata systems.
TUTORIAL DAY
Gerhard Stoll and Russell Mason, papers cochairs, have
targeted a number of papers for tutorial presentations on
Thursday, June 17 as a good way to offer attendees a
thorough introduction to the subject of metadata. Two
invited papers in the first morning session, “Metadata,
Identities, and Handling Strategies,” by Chris
Chambers, and “Before There Was Metadata,” by Mark
Yonge, are introductory papers to set the stage for
everything that follows.
The next session, File Basics, has three invited
papers: “Introduction to MXF and AAF,” by Philip
DeNier; “XML Primer,” by Claude Seyrat; and “Keep-
ing it Simple: BWF and AES31,” by John Emmett.
The first session on Thursday afternoon, Practical
Schemes, starts with an invited paper by Philippa Mor-
rell, “The Role of Registries.” The next paper, by
researchers from Pompeu Fabra University of
Barcelona, will look at a system for managing sound
effects. And Richard Wright will present an invited
paper on the Dublin Core. The final session on Thurs-
day is a workshop on MPEG-7. This tutorial day is also
available as a single-day registration option (see the
registration form on page 411).
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April
 Metadata for Audio
London, UK
June 17–19, 2004
CONFERENCE DAY 1
On Friday the papers sessions begin with Frameworks, which
includes an invited paper by Wes Curtis, “P-META: Program
Data Exchange in Practice.” This will be followed by the first
posters session. And the final morning session will be
Toolkits, which will include two invited papers: “Digital
Media Project,” by R. Nicol, and “MPEG-21: What and
Why,” by Jan Bormans and Kate Grant. After lunch there will
be the two-part session Feature Extraction. The second part of
the posters session will also be on Friday afternoon.
On Friday evening there will be an optional (not included in
conference registration fee) banquet and guided tour at the his-
toric Houses of Parliament (see photo below). The evening
will start with a tour of the debating chambers of the Houses
of Commons and Lords. There will be a brief technical talk
about the sound-reinforcement system in the Lords Chamber,
which uses 84 microphones on motorized winches and has
over 400 individually controlled loudspeakers. Afterwards
dinner will be served in a room overlooking the River Thames.
CONFERENCE DAY 2
The entire Saturday morning session will be Broadcast
Implementations, which will include papers on the metadata
processes of British, German, and Japanese broadcasters.
To
AES
25th
J. Audio Eng. Soc., Vol. 52, No. 4, 2004
Metadata is the “bread and butter” of libraries
and archives, so the first afternoon session on
Saturday will include papers about projects at
the U.S. Library of Congress, Spanish National
Radio, and Swedish Radio. The final
conference session will be on metadata
needed for the online delivery of audio. The
calendar, complete program with abstracts,
and conference registration form follow on
pages 404–411.
And, of course, you should try to
schedule an extra day or two to visit one
of the world’s great cities. London
preserves its magnificent history and
at the same time encourages new
music, art, literature, and architec-
ture; just what good metadata
does for audio. Meet your col-
leagues there June 17–19 for the
AES 25th International Confer-
ence, Metadata for Audio, it’s
going to be absolutely fabulous.
Go to www.aes.org now for
more details and register online.
 AES 25th INTERNATIONAL CONFERENCE
Metadata for Audio
London, UK • June 17–19, 2004

THURSDAY, JUNE 17 FRIDAY, JUNE 18 SATURDAY, JUNE 19

TUTORIAL DAY CONFERENCE DAY 1 CONFERENCE DAY 2

Tutorial Registration Conference Registration

Session 1 Session 7
Session T-1 Broadcast Implementations, Session A
Introduction Frameworks

Session 2: Posters, Part 1 Session 8

Session T-2 Broadcast Implementations, Session B
File Basics Session 3
Toolkits
Lunch Lunch
Lunch

Session T-3 Session 9

Practical Schemes Session 4 Libraries and Archives
Feature Extraction, Session A
Session 10
Workshop
Session 5: Posters, Part 2 Delivery of Audio
MPEG-7
Session 6
Photo courtesy British Tourist Office

Feature Extraction, Session B

Banquet (optional)
Houses of Parliament

Schedule is subject to change. Check www.aes.org for updates.

 AES 25th International
Conference Program
Metadata for Audio
2004 June 17–19
London, UK
Technical Sessions
Thursday, June 17
TUTORIALS
SESSION T-1: INTRODUCTION
T1-1 Metadata, Identities, and Handling Strategies—
Chris Chambers, BBC R&D, Tadworth, Surrey, UK
(invited)
With all the potential media material and its associated
metadata becoming accessible on IT-based systems,
how are systems going to find and associate the ele-
ments of any single item? How are the users going to
know they have the correct items when assembling
audio, video, and information for use within a larger
project? This short talk will explore the way areas of
our industry are hoping to tackle the problem and
some of the standards being introduced to ensure
management of this material is possible.
T1-2 Before There Was Metadata—Mark Yonge (invited)
Audio has never existed in isolation. There has always
been a mass of associated information, both explicit
and implicit, to direct, inform, and enhance the use of
the audio. In the blithe days before information theory
we didn’t know it was all metadata. This paper reviews
the extent of traditional metadata covering a range of
forms. Some of them may be surprising; all of them
need to be re-appraised in the light of newer, more for-
mal metadata schemes.
Thursday, June 17
SESSION T-2: FILE BASICS
T2-1 Introduction to MXF and AAF—Philip DeNier, BBC
R&D, Tadworth, Surrey, UK (invited)
The AAF and MXF file formats provide a means to
exchange digital media along with a rich (extendible)
set of metadata. This presentation will be a basic intro-
duction into the content of these file formats and will
include a description of the metadata scheme used.
T2-2 XML Primer—Claude Seyrat (invited)
Most audio professionals have heard of the term “XML”
but not many know for sure what it means or have yet
had to work with it. This paper sets out what XML is,
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April
what it can do for the user, and various ways that it
can be employed in the area of metadata for
audio. The contents of this paper form the basis for a
number of the papers that appear later in the conference.
T2-3 Keeping it Simple: BWF and AES31—John Emmett,
Broadcast Project Research Ltd., Teddington,
Middlesex, UK (invited)
Digital audio is spreading outward to the furthest
reaches of the broadcast chain. Making the best use of
the opportunities presented by this demands a stan-
dardization procedure that is adaptable to a vast num-
ber of past, present, and future digital audio formats
and scenarios. In addition, would it not be just great if
it cost nothing? This paper will point out the benefits of
what we already have and tell a tale of borrowing
economical audio technology from many sources.
Thursday, June 17
SESSION T-3: PRACTICAL SCHEMES
T3-1 The Role of Registries—Phillipa Morell, Metadata
Associates Ltd., London, UK (invited)
Some forms of metadata, especially those that identify
objects or classes of objects, form classes of their own
that need to be administered centrally in order to avoid
the risk of duplication and consequent misidentifica-
tion. The concept of such a registry is not new; for
example, International Standard Book Numbers (ISBN)
derive from a central registry that was originally set up
in 1970. The registry that ensures that every ethernet-
connected device in the world is uniquely identifiable is
another example. Formal identifiers and other metada-
ta for use in commercial transactions will increasingly
use the services of one or more metadata registries, as
this paper will discuss.
T3-2 Sound Effect Taxonomy Management in
Production Environments—Pedro Cano, Markus
Koppenberger, Perfecto Herrera, Oscar Celma,
Universitat Pompeu Fabra, Barcelona, Spain
Categories or classification schemes offer ways of nav-
igating and having higher control over the search and
retrieval of audio content. The MPEG-7 standard pro-
vides description mechanisms and ontology man- ➥
405
25rd International Conference Program
agement tools for multimedia documents. We have
implemented a classification scheme for sound effects
management inspired by the MPEG-7 standard on top
of an existing lexical network, WordNet. WordNet is a
semantic network that organizes over 100,000 con-
cepts of the real world with links between them. We
show how to extend WordNet with the concepts of the
specific domain of sound effects. We review some of
the taxonomies to acoustically describe sounds. Mining
legacy metadata from sound effects libraries further
supplies us with terms. The extended semantic net-
work includes the semantic, perceptual, and sound
effects specific terms in an unambiguous way. We
show the usefulness of the approach, easing the task
for the librarian and providing higher control on the
search and retrieval for the user.
T3-3 Dublin Core—R. Wright, BBC (invited)
Dublin Core metadata provides card catalog-like defini-
tions for defining the properties of objects for Web-
based resource discovery systems. The importance of
the Dublin Core is its adoption as a basis for many
more elaborate schemes. When the view ahead is
obscured by masses of local detail, a firm grasp of the
Dublin Core will often reveal the real landscape.
Thursday, June 17
WORKSHOP—MPEG-7
Coordinator: G. Peeters, IRCAM, Paris, France
(in association with SAA TC)
Managing Large Sound Databases Using MPEG—
Max Jacob, IRCAM, Paris, France
Sound databases are widely used for scientific, com-
mercial, and artistic purposes. Nevertheless there is yet
no standard way to manage them. This is due to the
complexity of describing and indexing audio content and
to the variety of purposes a sound database might
address. Recently there appeared MPEG-7, a standard
for audio/visual content metadata that could be a good
starting point. MPEG-7 not only defines a set of descrip-
tion tools but is more generally an open framework host-
ing specific extensions for specific needs in a common
environment. This is crucial since there would be no way
to freeze in a monolithic definition all the possible needs
of a sound database. This paper outlines how the
MPEG-7 framework can be used, how it can be extend-
ed, and how all this can fit into an extensible database
design, gathering three years of experience during the
CUIDADO project at IRCAM.
Integrating Low-Level Metadata in Multimedia
Database Management Systems—Michael Casey,
City University, London, UK
[Abstract Not Available at Press Time]
Tools for Content-Based Retrieval and
Transformation of Audio Using MPEG-7: The
SPOff and the MDTools—Emilia Gómez, Oscar
Celma, Emilia Gómez, Fabien Gouyon, Perfecto
Herrera, Jordi Janer, David García, University
Pompeu Fabra, Barcelona, Spain
In this workshop we will demonstrate three applica-
tions for content-based retrieval and transformations
of audio recordings. They illustrate diverse aspects of
a common framework for music content description
and structuring implemented using the MPEG-7 stan-
dard. MPEG-7 descriptions can be generated either
406
manually or automatically and are stored in an XML
database. Retrieval services are implemented in the
database. A set of musical transformations are
defined directly at the level of musically meaningful
MPEG-7 descriptors and are automatically mapped
onto low-level audio signal transformations. Topics
included in the presentation are: (1) Description gen-
eration procedure, manual annotation of editorial
description: the MDTools, automatic description of
audio recordings, the SPOffline; (2) Retrieval function-
alities, local retrieval: SPOffline, remote retrieval:
Web-based retrieval; and (3) Transformation utilities:
the SPOffline.
Using MPEG-7 Audio Low-Level Scalability: A
Guided Tour—Jürgen Herre, Eric Allamanche,
Fraunhofer IIS, Ilmenau, Germany
[Abstract Not Available at Press Time]
Friday, June 18
CONFERENCE DAY 1
SESSION CD-1: FRAMEWORKS
1-1 Data Model for Audio/Video Production—A. Ebner,
IRT, Munich, Germany
When changing from traditional production systems to
IT-based production systems the introduction and
usage of metadata is unavoidable. Direct access of the
information stored in IT-based systems is not possible.
Descriptive and structural metadata are the enablers to
have proper access of selected material. Metadata
does not focus on descriptive information about the
content only. It describes the usage of the material, the
structure of a program, handling processes, relevant
information, delivery information about properties, and
storage of information. The basis to achieve a com-
plete collection of metadata is a detailed analysis of a
broadcaster's production processes and usage cases.
A logical data model expresses the relationship
between the information and is the foundation for
implementations that enable a controlled exchange
and storage of metadata.
1-2 P-META: Program Data Exchange in Practice—
Wes Curtis, BBC Television, London, UK (invited)
[Abstract Not Available at Press Time]
Friday, June 18
SESSION CD-2: POSTERS, PART 1
2-1 Low-Complexity Musical Meter Estimation from
Polyphonic Music—Christian Uhle1, Jan Rohden1,
Markus Cremer1, Jürgen Herre2
1Fraunhofer AEMT, Erlangen, Germany
2Fraunhofer IIS, Ilmenau, Germany
This paper addresses the automated extraction of
musical meter from audio signals on three hierarchical
levels, namely tempo, tatum, and measure length. The
presented approach analyzes consecutive segments
of the audio signal equivalent to a few seconds length
each, and detects periodicities in the temporal progres-
sion of the amplitude envelope in a range between
0.25 Hz and 10 Hz. The tatum period, beat period, and
measure length are estimated in a probabilistic manner
from the periodicity function. The special advantages
of the presented method reside in the ability to track
tempo also in music with strong syncopated rhythms,
and its computational efficiency.
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April
25rd International Conference Program
2-2 Percussion-Related Semantic Descriptors of
Music Audio Files—Perfecto Herrera1, Vegard
Sandvold2, Fabien Gouyon1
1Universitat Pompeu Fabra, Barcelona, Spain
2University of Oslo, Oslo, Sweden
Automatic extraction of semantic music content meta-
data from polyphonic audio files has traditionally
focused on melodic, rhythmic, and harmonic aspects.
In the present paper we will present several music con-
tent descriptors that are related to percussion instru-
mentation. The “percussion index” estimates the
amount of percussion that can be found in a music
audio file and yields a (numerical or categorical) value
that represents the amount of percussion detected in
the file. A further refinement is the “percussion profile,”
which roughly indicates the existing balance between
drums and cymbals. We finally present the “percusive-
ness” descriptor, which represents the overall impul-
siveness or abruptness of the percussive events. Data
from initial evaluations, both objective (i.e., errors,
misses, false alarms) and subjective (usability, useful-
ness) will also be presented and discussed.
2-3 Tonal Description of Polyphonic Audio for Music
Content Processing—Emilia Gómez, Perfecto
Herrera, Universitat Pompeu Fabra, Barcelona, Spain
The purpose of this paper is to describe a system that
automatically extracts metadata from polyphonic audio
signals. This metadata describes the tonal aspects of
music. We use a set of features to estimate the key of
the piece and to represent its tonal structure, but they
could also be used to measure the tonal similarity
between two songs and to perform some key-based
segmentation or establish a tonal structure of a piece.
2-4 Phone-Based Spoken Document Retrieval in
Conformance with the MPEG-7 Standard—Nicolas
Moreau, Hyoung-Gook Kim, Thomas Sikora,
Technical University of Berlin, Berlin, Germany
This paper presents a phone-based approach of spo-
ken document retrieval, developed in the framework of
the emerging MPEG-7 standard. The audio part of
MPEG-7 encloses a SpokenContent tool that provides
a standardized description of the content of spoken
documents. In the context of MPEG-7, we propose an
indexing and retrieval method that uses phonetic infor-
mation only and a vector space IR model. Experiments
are conducted on a database of German spoken docu-
ments with ten city name queries. Two phone-based
retrieval approaches are presented and combined. The
first one is based on the combination of phone
N-grams of different lengths used as indexing terms.
The other consists of expanding the document repre-
sentation thanks to the phone confusion probabilities.
2-5 Efficient Features for Musical Instrument
Recognition on Solo Performances—Slim Essid,
Gaël Richard, Bertrand David, GET-Télécom Paris
(ENST), Paris, France
Musical instrument recognition is one of the important
goals of musical signal indexing. While much effort has
already been dedicated to such a task, most studies
were based on limited amounts of data that often
included only isolated musical notes. In this paper we
address musical instrument recognition on real solo
performance based on larger training and test sets. A
highly efficient set of features is proposed that is
obtained from signal cepstrum but also from spectrum
low- and higher-order statistical moments describing
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April
signal spectral shape. The use of principal component
analysis in conjunction with support vector machine
classification yields to nearly perfect recognition accu-
racy on varied musical solo phrases from ten instru-
ments issued from different instrument families.
Friday, June 18
SESSION CD-3: TOOLKITS
3-1 Digital Media Project—R. Nicol, BT, Ipswich, UK
(invited)
[Abstract Not Available at Press Time]
3-2 MPEG-21: What and Why—Jan Bormans1, Kate
Grant2 (invited)
1IMEC, Leuven, Belgium
2Nine Tiles, Cambridge, UK
The MPEG-21 vision is to define a multimedia frame-
work to enable transparent and augmented use of mul-
timedia resources across a wide range of networks
and devices used by different communities. The tech-
nical report “Vision, Technologies and Strategy”
describes the two basic building blocks: the definition
of a fundamental unit of distribution and transaction
(the digital item) and the concept of users interacting
with digital items. The digital items can be considered
the “what” of the multimedia framework (e.g., a video
collection, a music album), and the users can be con-
sidered the “who” of the multimedia framework. MPEG-
21 is developing a number of specifications
enabling the integration of components and standards
to facilitate harmonisation of “technologies” for the cre-
ation, modification, management, transport, manipula-
tion, distribution, and consumption of digital items. This
paper will explain the relationship of the different
MPEG-21 specifications by describing a detailed use-
case scenario.
3-3 A 3-D Audio Scene Description Scheme Based on
XML—Guillaume Potard, Ian Burnett, University of
Wollongong, NSW, Australia
An object-oriented schema for describing time-vary-
ing 3-D audio scenes is proposed. The creation of
this schema was motivated by the fact that current
virtual reality description schemes (VRLM, X3D) have
only basic 3-D audio description capabilities. In con-
trast, MPEG-4 AudioBIFs have advanced 3-D audio
features but are not designed as a metadata lan-
guage. MPEG-4 BIFs are particularly targeted as a
binary scene description language for scene render-
ing purposes only. Our proposed 3-D audio scene
description schema features state-of-the art 3-D
audio description capabilities while being usable both
as a metadata scheme for describing 3-D audio con-
tent (for example, 5.1 or Ambisonics B-format) and as
a format for scene rendering.
Friday, June 18
SESSION CD-4: FEATURE EXTRACTION, SESSION A
4-1 A System for Harmonic Analysis of Polyphonic
Music—Claas Derboven, Markus Cremer, Fraunhofer
IIS AEMT, Ilmenau, Germany
A system for harmonic analysis of polyphonic musical
signals is presented. The system uses a transform with
a nonuniform frequency resolution for the extraction of
prominent tonal components and determines the key
and the contained chords of a musical input signal with
high accuracy. A statistical approach based on the ➥
407
25rd International Conference Program
frequency of occurrence of musical notes for determin-
ing the key is described. An algorithmic solution for
chord determination is presented with a concise expla-
nation. Finally, a qualitative evaluation of the system’s
performance is conducted to demonstrate the applica-
bility to real-world audio signals.
4-2 Robust Identification of Time-Scaled Audio—Rolf
Bardeli, Frank Kurth, University of Bonn, Bonn, Germany
Automatic identification of audio titles on radio broad-
casts is a first step toward automatic annotation of
radio programs. Systems designed for the purpose of
identification have to deal with a variety of postpro-
cessing potentially imposed on audio material at the
radio stations. One of the more difficult techniques to
be handled is time-scaling, i.e., the variation of play-
back speed. In this paper we propose a robust finger-
printing technique designed for the identification of
time-scaled audio data. This technique has been
applied as a feature extractor to an algebraic indexing
technique that has already been successfully applied
to the task of audio identification.
4-3 Computing Structural Descriptions of Music
through the Identification of Representative
Excerpts from Audio Files—Bee Suan Ong,
Perfecto Herrera, Universitat Pompeu Fabra,
Barcelona, Spain
With the rapid growth of audio databases, many music
retrieval applications have employed metadata
descriptions to facilitate better handling of huge data-
bases. Music structure creates the uniqueness identity
for each musical piece. Therefore, structural descrip-
tion is capable of providing a powerful way of interact-
ing with audio content and serves as a link between
low-level description and higher-level descriptions of
audio (e.g., audio summarization, audio fingerprinting,
etc.). Identification of representative musical excerpts
is the primary step toward the goal of generating struc-
tural descriptions of audio signals. In this paper we dis-
cuss various approaches in identifying representative
musical excerpts of music audio signals and propose
to classify them into a few categories. Pros and cons of
each approach will also be discussed.
Friday, June 18
SESSION CD-5: POSTERS, PART 2
5-1 Toward Describing Perceived Complexity of
Songs: Computational Methods and
Implementation—Sebastian Streich, Perfecto
Herrera, Universitat Pompeu Fabra, Barcelona, Spain
Providing valuable semantic descriptors of multimedia
content is a topic of high interest in current research.
Such descriptors should merge the two predicates of
being useful for retrieval and being automatically
extractable from the source. In this paper the semantic
descriptor concept of music complexity is introduced. Its
benefit for music retrieval and automated music recom-
mendation is addressed. The authors provide a critical
review of existing methods and a detailed prospect of
new methods for automated music complexity estimation.
5-2 How Efficient Is MPEG-7 for General Sound
Recognition?—Hyoung-Gook Kim, Juan José
Burred, Thomas Sikora, Technical University Berlin,
Berlin, Germany
Our challenge is to analyze/classify video sound track
content for indexing purposes. To this end we compare
408
the performance of MPEG-7 audio spectrum projection
(ASP) features based on several basis decomposition
algorithms vs. mel-scale frequency cepstrum coeffi-
cients (MFCC). For basis decomposition in the feature
extraction we evaluate three approaches: principal com-
ponent analysis (PCA), independent component analy-
sis (ICA), and non-negative matrix factorization (NMF).
Audio features are computed from these reduced vec-
tors and are fed into a hidden Markov model (HMM)
classifier. We found that established MFCC features
yield better performance compared to MPEG-7 ASP in
general sound recognition under practical constraints.
5-3 Automatic Optimization of a Musical Similarity
Metric Using Similarity Pairs—Thorsten Kastner,
Eric Allamanche, Oliver Hellmuth, Christian Ertel,
Marion Schalek, Jürgen Herre, Fraunhofer IIS,
Ilmenau, Germany
With the growing amount of multimedia data available
everywhere and the necessity to provide efficient meth-
ods for browsing and indexing this plethora of
audio content, automated musical similarity search and
retrieval has gained considerable attention in recent
years. We present a system which combines a set of
perceptual low-level features with appropriate classifica-
tion schemes for the task of retrieving similar sounding
songs in a database. A methodology for analyzing the
classification results to avoid time consuming subjective
listening tests for an optimum feature selection and com-
bination is shown. It is based on a calculated “similarity
index” that reflects the similarity between specifically em-
bedded similarity pairs. The system’s performance as
well as the usefulness of the analyzing methodology is
evaluated through a subjective listening test.
5-4 Automatic Extraction of MPEG-7 Metadata for
Audio Using the Media Asset Management System
iFinder—Jobst Löffler, Joachim Köhler, Fraunhofer
IMK, Sankt Augustin, Germany
This paper describes the MPEG-7 compliant media
asset management system iFinder, which provides a
set of automatic methods and software tools for media
analysis, archiving, and retrieval. The core technology
of iFinder comprises several modules for audio and
video metadata extraction that are bundled in the
iFinderSDK, a commercial product offered to the media
industry. The workflow for audio content processing
together with pattern recognition methods used will be
presented. Of special note, a technique for precise
audio-text alignment together with a browser applica-
tion for synchronized display of retrieval results will be
demonstrated. An insight to using MPEG-7 as a stan-
dardized metadata format for media asset manage-
ment will be provided from a practical point of view.
5-5 An Opera Information System Based on MPEG-7—
Oscar Celma Herrada, Enric Mieza, Universitat
Pompeu Fabra, Barcelona, Spain
We present an implementation of the MPEG-7 stan-
dard for a multimedia content description of lyric opera
in the context of the European IST project: OpenDra-
ma. The project goals are the definition, development,
and integration of a novel platform to author and deliv-
er the rich cross-media digital objects of lyric opera.
MPEG-7 has been used in OpenDrama as the base
technology for a music information retrieval system. In
addition to the MPEG-7 multimedia description
scheme, different classification schemes have been
proposed to deal with operatic concepts such as musi-
cal forms (acts, scenes, frames, introduction, etc.),
J. Audio Eng. Soc., Vol. 52, No. 3, 2004 April
25rd International Conference Program
musical indications (piano, forte, ritardando, etc.), and
genre and creator roles (singers, musicians, production
staff, etc.). Moreover, this project has covered the
development of an authoring tool for an MPEG-7 stan-
dard, namely MDTools, which includes segmentation,
classification scheme generation, creation and produc-
tion, and media information descriptors.
5-6 Morphological Sound Description: Computational
Model and Usability Evaluation—Julien Ricard,
Perfecto Herrera, Universitat Pompeu Fabra,
Barcelona, Spain
Sound samples of metadata are usually limited to low-
level descriptors and a source label. In the context of
sound retrieval only the latter is used as a search cri-
terion, which makes the retrieval of sounds having no
identifiable source (abstract sounds) a difficult task.
We propose a description framework focusing on
intrinsic perceptual sound qualities, based on Schaef-
fer’s research on sound objects that could be used to
represent and retrieve abstract sounds and to refine a
traditional search by source for nonabstract sounds.
We show that some perceptual labels can be auto-
matically extracted with good performance, avoiding
the time-consuming manual labeling task, and that the
resulting representation is evaluated as useful and
usable by a pool of users.
Friday, June 18
SESSION CD-6: FEATURE EXTRACTION, SESSION B
6-1 Drum Pattern-Based Genre Classification from
Popular Music—Christian Uhle, Christian Dittmar,
Fraunhofer AEMT, Ilmenau, Germany
This paper addresses the identification of drum patterns
and the classification of their musical genres. The drum
patterns are estimated from audio data automatically.
This process involves the transcription of percussive
unpitched instruments with a method based on indepen-
dent subspace analysis and a robust estimation of the
tatum grid and the musical meter. The rhythmic patterns
are identified from pattern histograms, describing the
frequency of occurrence of the percussive events. The
classification procedure evaluates the meter information,
the pattern histogram as well as other high-level rhyth-
mic features derived from the estimated drum pattern.
6-2 Assessing the Relevance of Rhythmic Descriptors
in a Musical Genre Classification Task—Fabien
Gouyon1, Simon Dixon2, Elias Pampalk2, Gerhard
Widmer2
1Universitat Pompeu Fabra, Barcelona, Spain
2Austrian Research Institute for AI, Vienna, Austria
Organizing or browsing music collections in a musically
meaningful way calls for tagging the data in terms of, e.g.,
rhythmic, melodic or harmonic aspects, among others. In
some cases, such metadata can be extracted automati-
cally from musical files; in others, a trained listener must
extract it by hand. In this paper we consider a specific set
of rhythmic descriptors for which we provide procedures of
automatic extraction from audio signals. Evaluating the rel-
evance of such descriptors is a difficult task that can easily
become highly subjective. To avoid this pitfall, we as-
sessed the relevance of these descriptors by measuring
their rate of success in genre classification experiments.
6-3 Music Genre Estimation from Low-Level Audio
Features—Oliver Hellmuth, Eric Allamanche,
Thorsten Kastner, Ralf Wistorf, Nicolas Lefebvre,
Jürgen Herre, Fraunhofer IIS, Ilmenau, Germany
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April
Despite the subjective nature of associating a certain
song or artist with a specific musical genre, this type
of characterization is frequently used to provide a con-
venient way of expressing very coarse information on
the basic stylistic and rhythmic elements and/or instru-
mentation of a song. An audio database that is struc-
tured according to different musical genres is a first
important step to provide an easy/intuitive access to a
large music collection. Thus, a convenient way for
indexing large databases by musical genre is desired.
This paper describes a system for an automatic genre
classification into several musical genres. Different
features as well as classification strategies will be
evaluated and compared. The system’s performance
is assessed by means of a subjective listening test.
Saturday, June 19
Conference Day 2
SESSION CD-7: BROADCAST IMPLEMENTATIONS,
SESSION A
7-1 Audio Metadata in Radio Broadcasting— Shigeru
Aoki1, Masahito Kawamori2
1TokyoFM Broadcasting, Tokyo, Japan
2NTT, Tokyo, Japan
Generally an audio sequence or program is produced by
DAW (Digital Audio Workstation) and delivered as a dig-
ital audio file. However, the descriptive data of the audio
program, such as the cue sheet of the radio program, is
transferred apart from the audio file. The content
descriptive data is commonly known as metadata. The
most effective method to transfer the audio data and the
metadata is to embed those as one digital file that an
audio player plays and offer the description of that audio
sequence simultaneously. This paper describes the for-
mat and scheme of the audio file with metadata.
7-2 Integrated Metadata in the Broadcast Environment
—Joe Bull1, Kai-Uwe Kaup2
1SADiE UK, Cambridgeshire, UK
2VCS Aktiengesellschaft, Bochum, Germany
In a modern broadcast environment, efficient and effec-
tive handling of metadata becomes more important
every day. Much time and money can be wasted reen-
tering data that is already present in the digital domain.
This money could be better spent on program-making.
The authors will describe practical examples of how this
can be achieved in a real broadcast environment using
real products in use or in development.
Saturday, June 19
SESSION CD-8: BROADCAST IMPLEMENTATIONS
SESSION B
8-1 Broadcast Wave and AES Audio in MXF—Bruce
Devlin, Snell & Wilcox
The SMPTE has established MXF is the new open stan-
dard file format for interchange in the broadcast world.
One important aspect of the standard is audio mapping.
This paper will be a basic tutorial on how MXF and the
audio mapping standard work. It will include issues of
physically interleaving audio and video as well as adding
rich metadata using the MXF data model.
8-2 The Advanced Authoring Format and its Relevance
to the Exchange of Audio Editing Decisions—David
McLeish1, Phil Tudor2
1SADiE, Cambridgeshire, UK,
2BBC R&D, Tadworth, Surrey, UK
➥
409
25rd International Conference Program
This paper explores how the Advanced Authoring
Format (AAF) model, a vehicle for exchanging
metadata and media-rich content, can be used to
describe audio program compositions so that they
can be more seamlessly exchanged between audio
editors who work with tools designed by different
manufacturers. In addition, the extensibility of the
format is discussed as a means of looking at its fu-
ture potential.
Saturday, June 19
SESSION CD-9: LIBRARIES AND ARCHIVES
9-1 Development of a Digital Preservation Program at
the Library of Congress—Carl Fleischhauer,
Samuel Brylawski, Library of Congress, Washington,
DC, USA
This paper will trace the development of a digital
preservation program for sound recordings at the
Library of Congress. It will outline the Library’s use of
METS (Metadata Encoding and Transmission Stan-
dard); survey the challenges faced in the Library’s
work to create digital objects for public use, comprised
of sound files and images of packaging and accompa-
nying materials; and review the tools and methods
utilized to create metatdata.
9-2 Audio Metadata Used in the Radio Nacional de
España Sound Archive Project—Miguel Rodeno1,
Jesus Nicolas2, Isabel Diaz2
1Alcala University, Madrid, Spain
2Radio Nacional de España, Madrid, Spain
The 20th century Spanish sound history has been pre-
served in digital format and may now be consulted
online through the Internet. This a pioneer project in the
broadcasting industry around the world finished in
December 2002. The archive is considered the most
important audio archive in the Spanish language in the
world. This paper describes the metadata used in this
project. Radio Nacional de España followed the 1997/98
European Broadcasting Union (EBU) standard for the
interchange of audio files and their broadcasting: the
Broadcast Wave Fornmat (BWF). The voice/word and
the different kinds of music (classical, light, international
or Spanish) have different types of metadata. Some
examples are shown with the detailed metadata.
9-3 Integration of Audio Computer Systems
and Archives Via the SAM/EBU Dublin Core
Standard,Tech.doc 3293—Lars Jonsson1, Gunnar
Dahl2
1Swedish Radio
2KSAD, Norsk Rikskringkasting, Oslo, Sweden
Dublin Core is a well-known metadata initiative from
W3C that has been widely spread and used for text
and Web pages on the Internet. The Scandinavian
SAM-group, with 25 archive specialists and engineers
have defined semantic definitions and converted the
commonly used Dublin Core initiative for general use
within the audio industry. The 15 basic elements of
Dublin Core and new subsets have proven to cover
most of the tape protocols and database fields existing
in broadcast production chain from early capturing over
various types of production and all the way to distribu-
tion and archiving. This presentation covers some
examples of the use of metadata transfer with Dublin
Core expressed in XML in Sweden and Norway. It
ends in a discussion of the future possibilities of Dublin
Core in comparison with other existing metadata initia-
410
tives in an integrated world of interconnected databas-
es coming into all audio-related companies.
Saturday, June 19
SESSION CD-10: DELIVERY OF AUDIO
10-1 Watermarking and Copy Protection by Information
Hiding in Soundtracks—Tim Jackson, Keith Yates,
Francis Li, Manchester Metropolitan University,
Manchester, UK
In this paper digital audio watermarking techniques
are reviewed and categorized. Applications of water-
marking schemes are discussed, and their capabili-
ties and limitations clarified in the context of audio
copyright management and copy protection. Tradi-
tional watermarking schemes embed hidden signa-
tures in soundtracks and are found to be effective in
ownership authentication and copyright manage-
ment. Nevertheless, they do not prevent unautho-
rized copying unless dedicated watermark detectors
are added to the recording devices. Purpose-chosen
hidden signals are known to interfere with some
recording devices, for example, magnetic tape
recorders, offering a potential solution to copy pro-
tection. It is therefore reasonable to postulate that
watermarking techniques could be extended to the
general audio copy protection without the resort to
dedicated detectors.
10-2 Metadata Requirements for Enabling the On-line
Music Industry’s New Business Models and
Pricing Structures—Nicolas Sincaglia, MusicNow
Inc., Chicago, IL, USA
The music industry has begun selling and distributing its
media assets online. Online music distribution is vastly
different from the normal means of media distribution.
These untested methods of music sales and distribution
require experimentation in order to determine which
business models and pricing tiers will most resonate
with the consumer. This translates into the need for ver-
satile and robust data models to enable these market tri-
als. Copyright owners and media companies require
well-designed data structures to enable them to transmit
and receive these complicated sets of business rules.
This metadata is an essential part of an overall digital
rights management system to control and limit access to
the associated media assets.
10-3 Audio Meta Data Generation for the Continuous
Media Web—Claudia Schremmer1, Steve Cassidy2,
Silvia Pfeiffer1
1CSIRO, Epping, NSW, Australia
2Macquarie University, Sydney, Australia
The Continuous Media Web (CMWeb) integrates time-
continuous media into the searching, linking, and brows-
ing function of the World Wide Web. The file format un-
derlying the CMWeb technology, Annodex, streams the
media content multiplexed with metadata in CMML for-
mat that contains information relevant to the whole media
file (e.g., title, author, language) as well as time-sensitive
information (e.g., topics, speakers, time-sensitive hyper-
links). This paper discusses the problem of generating
Annodex streams from complex linguistic annotations:
annotated recordings collected for use in linguistic
research. We are particularly interested in
automatically annotated recordings of meetings and tele-
conferences and see automatically-generated CMML
files as one way of viewing such recordings. The paper
presents some experiments with generating Annodex
files from hand annotated meeting recordings.
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April
 AES 25TH INTERNATIONAL CONFERENCE REGISTRATION
2004 June 17–19, London, UK
Registration: Fee includes attendance at the 25th minutes walk from Church House. Delegates will
conference including a copy of the Proceedings of have the opportunity to continue their discussions
the AES 25th International Conference as well as there in a relaxed atmosphere. There are three
lunch and refreshments Thursday through price options—£99, £119, and £149—with VAT and
Saturday. There is an optional dinner on Friday breakfast included; these are special rates for the
evening at the Houses of Parliament. Rooms have conference. To book a room fill out item 2 below;
been reserved at St. Ermin’s Hotel, which is a few payment is made to hotel at end of stay.
Please return by mail or fax to:
Heather Lane, AES, P.O. Box 645, Slough SL1 8BJ, UK
Fax: +44 1628 667002 Tel: +44 1628 663725 Email: uk@aes.org
Conference registration is also available online at www.aes.org/events/25/registration.html.
No registration is final until the conference fee has been received.

Please print/type all information as you wish it to appear on your badge.

1
First Name Last Name
Affiliation (Company/Institution)
Street Address
City State Postal Code Country
Telephone INT+ Fax INT+
Email AES Membership No.

■ Book me a room at St. Ermin’s Hotel Number of people______

2 ■ £ 99 Standard ■ £ 119 Superior ■ £ 149 Deluxe (Prices are per room, not per person)
Arrival Date Departure Date Number of nights

Metadata Conference only (Friday and Saturday, June 18–19)

3 ■ AES MEMBERS £ 298.00 plus £52.15 VAT Total: £ 350.15
■ NONMEMBERS £ 348.00 plus £60.90 VAT Total: £ 408.90
■ AUTHORS, STUDENTS £ 198.00 plus £34.65 VAT Total: £ 232.65
Tutorial Day only (Thursday, June 17)
■ AES MEMBERS £ 98.00 plus £17.15 VAT Total: £ 115.15
■ NONMEMBERS £ 128.00 plus £22.40 VAT Total: £ 150.40
■ AUTHORS, STUDENTS £ 68.00 plus £11.90 VAT Total: £ 79.90
Tutorial Day and Conference
■ AES MEMBERS £ 368.00 plus £64.40 VAT Total: £ 432.40
■ NONMEMBERS £ 448.00 plus £78.40 VAT Total: £ 526.40
■ AUTHORS, STUDENTS £ 248.00 plus £43.40 VAT Total: £ 291.40
■ Optional Dinner___number of tickets £ 38.00 plus £ 6.65 VAT Total: £ 44.65

Payment Modes (check box) Total Amount £_________________________

4
■ Check (payable to AES Ltd.) ■ Mastercard/Eurocard ■ Visa
Card Number Expiration Date

/
Month Year
Cardholder's Name (print)________________________________________________________________________
Signature of Cardholder__________________________________________________________________________

J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April 411

 HISTORICAL
PERSPECTIVES AND
TECHNOLOGY OVERVIEW
OF LOUDSPEAKERS
FOR SOUND
REINFORCEMENT*
1,2
J. Eargle , Honorary Member, and M. Gander , AES Fellow
(1) JBL Professional, Northridge, California, U. S. A.
(2) JME Consulting Corporation, Los Angeles, California, U. S. A.
INTRODUCTION
Horns and direct-radiating systems
have provided the basis for sound rein-
forcement for more than a century.
Both technologies have benefited from
engineering and manufacturing im-
provements as well as demands for
pushing the performance envelope.
Trends of fashion have often inter-
sected with engineering development,
economics, and even marketplace
opportunism. A survey tutorial of the
significant developments in transduc-
tion, signal transmission, and system
synthesis is presented here and dis-
cussed in historical perspective.
We begin with an overview of sound
reinforcement and the technologies
that have supported it. This is followed
by more detailed technical discussions
of both direct radiating and horn sys-
tems, leading to a discussion of mod-
ern loudspeaker array techniques. The
presentation ends with a comprehen-
sive bibliography.
HISTORICAL PERSPECTIVES
In the early days of transducer develop-
ment, horn systems offered the only
means possible for achieving suitable
*Revised and expanded from a presenta-
tion at the Institute of Acoustics 12th
Annual Weekend Conference, Winder-
mere, England, October 25-27, 1996.
412
levels for speech reinforcement. Early
power amplifiers were limited to about
10 watts output capability, and horn-
driver efficiencies on the order of 20%
to 30% were necessary to reach the
desired sound pressure levels.
The direct field level referred to a
distance of one meter produced by one
acoustic watt radiated omnidirection-
ally from a point source in free space is
109 dB LP [1, p. 314]. If we use a 10-
watt amplifier with a horn–driver com-
bination that is 30% efficient, we can
produce three acoustical watts. If the
horn has a directivity index (DI) on
axis of, say, 10 dB, then we can
increase that level to:
Level (re 1 meter) = 109 +
10 log (3) + DI = 124 dB LP
At a more realistic listening distance
of 10 meters, the level would be, by
inverse square relationship, 20 dB
lower, or 104 dB. If wider coverage is
needed, more horns can be added and
splayed as required.
There is little documentation of early
examples of general speech reinforce-
ment, and that art progressed fairly
slowly [9]. The first example of large-
scale sound reinforcement occurred on
Christmas Eve, 1915, when E. S. Prid-
ham, cofounder of the Magnavox com-
1
pany, played Christmas carols for an
audience of 50,000 using Pridham-
Jensen rocking armature transducers
connected to phonograph horns [12].
Western Electric set up a public
address system capable of addressing
12,000 persons through 18 loudspeak-
ers in 1916 [24, p. 24].
The first distributed system was
employed in 1919; 113 balanced arma-
ture driving units mounted on long
horns were strung along New York
City’s Park Avenue “Victory Way” as
a part of a Victory Bond sale [24, p. 25]
[2], as shown in Fig. 1. The first suc-
cessful indoor use of a public address
system was at the 1920 Chicago
Republican Convention, which also
employed the first central cluster con-
figuration [24, p. 25], as shown in Fig.
2. On March 4, 1921, President Hard-
ing’s inauguration was amplified [24,
p. 24], and on November 11, 1921,
President Harding’s address in Arling-
ton, Virginia, was transmitted by West-
ern Electric, using Edgerton’s 1918
design of four-air-gap balanced-arma-
ture units. For the first time 150,000
people, at Madison Square Garden in
New York, in the adjoining park, and in
the Civic Auditorium in San Francisco,
simultaneously listened to a person
speaking [2].
It was the cinema that paved the way
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April
Fig. 1. First use of a distributed sound system,
New York, Park Avenue (1919). Note trumpet
horns in upper left hung above crowd aiming
downward. (Photo from Lester Cowan
Recording Sound for Motion Pictures, McGraw
Hill, New York, 1931.)
for rapid development of professional
sound reproduction. Talking pictures
required more than speech intelligibil-
ity, however. The single horns of the
day, as shown in Fig. 3, were limited
in response to the range from about
125 Hz to 3 kHz. While this was ade-
quate for speech, music required
greater bandwidth, especially at lower
frequencies [7].
The first widely accepted cinema sys-
tem used a two-way approach. A multi-
cellular high-frequency horn with a
compression driver was coupled to a
cone-driven, folded low-frequency horn
assembly [3] as shown in Fig. 4. Its
high level of performance and approval
by the Academy of Motion Picture Arts
and Sciences, in the form of a technical
achievement award, led to its accep-
tance as an industry standard [11].
The next step in low frequency (LF)
response was to employ multiple large-
diameter direct radiators in some sort of
“directional baffle” that could both load
the drivers for increased efficiency and
give them added forward directionality
[20]. The best LF direct-radiating trans-
ducers of the day were about 10 dB
lower in efficiency than horn systems,
but their use was mandated due to the
size and complexity of traditional bass
horns. When employed in multiples,
they could provide sufficient level and
coverage in large theaters. In time, the
various directional baffles evolved into
the familiar front-loaded quasi-horns
that became universally identified with
the cinema [16], as shown in Fig. 5.
Field coil energized transducers also
gave way to permanent magnets fol-
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April
Fig. 2. First central array sound system, 1920
Chicago Republican Convention. (Photo Fig. 3. Western Electric
from Frederick Thrasher, ed., Okay for “Roxy” horn type 12A
Sound … How the Screen Found its Voice, with 555W dynamic
Duell, Sloan, and Pearce, New York, 1946.) driver. (Photo, Cowan,
p. 329.)
lowing WWII [15].
The LF transducers
chosen for this appli-
cation had relatively
small moving masses
and high (Bl) 2 /R E
ratios. This enabled
them to efficiently
handle the acoustical
load transformed by
the horn and maintain
good response up to
about 400 Hz, at which
point the high fre-
quency (HF) horn took
over. The HF horn Fig.4. Shearer MGM horn Fig.5. The Hilliard and
itself was undergoing system, which won the Lansing designed Altec-
modifications aimed at Academy of Motion Picture Lansing A-4 “Voice of
Arts and Sciences scientific
improving its direc- award in 1936. Contributors to the Theatre” system,
tional response. RCA this program included Douglas 1945. Utilizing a short
LF straight exponential
favored the radial Shearer, John Hilliard, James horn with an acoustical
horn, while Bell Labo- B. Lansing, Harry Olson, John path length equal to that
ratories favored the Volkmann, and William Snow. of the HF horn to avoid
(Photo from Research Council delay between the LF
multicellular device of AMPAS, Motion Picture
[27]. Bell Laboratories Sound Engineering, Van and HF sections.
(Photo, Altec-Lansing)
later pioneered the use Nostrand, New York, 1938.)
of the acoustical lens
in similar applications [8, 14]. sumer high fidelity during the mid-
By the mid-1940s, HF horns had 1950s and 1960s, reached the point
established their primacy in the fre- where beneficial tradeoffs could be
quency range above 500 Hz, while made among the three “eternal” loud-
hybrid LF horns, which relied on reflex speaker variables of size, efficiency,
loading of cones below about 100 Hz, and LF bandwidth extension. As early
dominated the range below 500 Hz. as the mid-1950s, consumers began to
These families of components formed enjoy relatively small sealed systems
the basis of postwar sound reinforce- that provided substantial LF output [25,
ment activities and maintained that 26]. Additional analysis and synthesis
position for nearly three decades. of ported enclosure design by Thiele
By the early 1970s available ampli- and Small [23], following the earlier
fier power, which had been steadily ris- work of Locanthi [17], Novak [19], and
ing through the introduction of con- Benson [31], led to the general ➥
413
 HISTORICAL PERSPECTIVES AND TECHNOLOGY OVERVIEW
OF LOUDSPEAKERS FOR SOUND REINFORCEMENT
flat power bandwidth over its operating
A B spectrum in the range from 40 to 400
Hz. It uses high-powered drivers and
can accommodate approximately four
times the amplifier power as the system
Fig. 6A, but by the mid-1970s the cost
per watt of amplifier power had dimin-
ished significantly.
For large-scale music reinforcement
there were other considerations. LF
horn systems had developed along lines
unique to the music industry, since
fairly high mid-bass directivity was a
prime concern. The older radial horns
were likewise favored for their direc-
tionality, which increased at HF. Typi-
Fig. 6. Large systems for speech reproduction. Design using pre-1970 components,
multicellular exponential horns with ported LF horns (A, J. Audio Eng. Soc., vol. 20, cal usage is shown in Fig. 7.
pp. 571); design using post-1970 components, constant directivity HF horns and The conservative motion picture
ported LF direct radiators (B, courtesy KMK Ltd.) industry continued with two-way solu-
tions but eventually traded in its hybrid
adoption of ported enclosures as a LF the previous generation of hardware bass horns for standard ported systems
building block in professional sound (theater-type LF hybrid horns and HF and retired the old multicellular HF
system design. multicellular horns), while the system horns of the 1930s [5]. The concept of
Through the use of Thiele-Small shown in Fig. 6B makes use of ported flat power response in system design
driver parameters, LF systems could be LF systems and modern HF horn hard- stated that not only uniform direct
rationally designed for best response ware. In terms of LF response and arrival sound was important; reflected
and enclosure size. Gone were the days power handling, both systems have six sound (proportional to the total radiated
of cutting and trying, and also gone LF drivers. With horn loading, the out- power of the system) should also be
were the days of poorly designed LF put capability of the system shown in uniform. This goal could be more eas-
transducers. Thiele-Small analysis Fig. 6A will be higher in the 100 to 200 ily met with ported LF systems, along
pointed to the need for identifying the Hz range than the system shown in Fig. with uniform coverage HF horns.
right cone mass, resonance, excursion 6B. However, the LF system in Fig. 6B Hilliard [10] was well ahead of his time
capability, Bl product, enclosure vol- is smaller and has been designed for in bringing this concept to the attention
ume, and tuning to achieve a targeted
performance goal.
During the 1970s and early 1980s
ported LF systems became the basis for
most sound reinforcement LF design.
For many music and speech applica-
tions the horn HF crossover frequency
was raised from 500 Hz to 800 Hz and
higher, with cone drivers filling in the
midrange [18]. This change from the
traditional two-way philosophy was
brought on by modern program
requirements for generating increased
level. Higher power handling transduc-
ers were substituted in the LF horns,
but these new drivers forced the HF
power bandwidth of the horn down-
ward. With the crossover frequency
raised to improve HF reliability, the
result was uniform on-axis response,
but grossly uneven power response [6].
By the mid-1970s ported LF systems
were poised to take a commanding lead
in the design of speech reinforcement
Fig. 7. Typical early-1970s horn-loaded system using ported LF horns, 90° radial
systems. Fig. 6 shows examples of the horns with large-format drivers, and 90° radial horns with small-format drivers for
older approach and the newer one. The augmenting the upper frequencies. A narrow coverage horn with a large format
system shown in Fig. 6A makes use of driver is used for long-throw coverage. (Photo, JBL)
414 J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April
 HISTORICAL PERSPECTIVES AND TECHNOLOGY OVERVIEW
OF LOUDSPEAKERS FOR SOUND REINFORCEMENT
of the film industry, but his attempts at As we continue our discussion of DIRECT RADIATORS
commercialization failed. direct radiators and horns, we will
The well-engineered ported LF sys- present a detailed analysis of the Early Development
tem of the 1970s established its domi- engineering fundamentals of both Ernst Werner Siemens’ 1874 U.S.
nance at the low end of the frequency methods of transduction. This dis- Patent 149,797 was prophetic; he
spectrum. The combination of low-cost cussion will cover basic operating described in detail a radial magnet struc-
power amplification, low-distortion, principles, directional control, and ture in which a coil of wire was placed.
high-power LF transducer development, distortion mechanisms associated The coil was connected with a radiating
and enclosure simplicity has given the with each method. We will follow surface that Siemens described as the
ported system an advantage at low fre- this with a discussion of array con- frustum of a cone. He had literally
quencies that all but the very largest cepts encompassing both kinds of invented the cone loudspeaker—with
horn systems could never match. transducers. nothing to play over it except dc tran- ➥
Keele [13] made an instructive com-
parison between LF horns and direct
radiators, pointing out the virtual parity
between them. For the same LF cutoff,
the horn will have highest efficiency, a TEST FASTER FOR LESS
large complex enclosure, and will
require a small number of drive units.
WITH DSCOPE SERIES III
By comparison, a direct radiator multi-
ple-driver vented box will have moder- Following completion of our extensive beta-test
ate efficiency, a small, relatively simple program, Release 1.00 is now available
enclosure, a large number of drivers,
and higher power handling capability
(because of the multiple drivers). In Ideal for:
Keele’s words, “This roughly means • Research & Development
that if one has a lot of space, not much
• Automated Production Test
money to spend on drivers and ampli-
fiers, and lots of cheap labor—build a • Quality Assurance
horn system. If labor is not cheap, and • Servicing
you don’t have much space, and if you • Installation
can afford drivers and amplifier
power—build a direct radiator system.”
In essence, the costs of labor and lum- DSNet I/O Switcher 16:2
ber had far outstripped those of drivers now available
and watts.
Through the late 1980s and 1990s,
however, the professional sound indus-
try saw a return to horn systems for
covering the range down to about 300
Hz, especially for large-scale speech
reinforcement. The reasons have to do
not with efficiency per se but rather
directional control. These new systems
have taken the form of large format
compression drivers, or cone transduc- dScope Series III is
ers designed for horn–driver applica- simply the fastest way to test.
tions, and large horns optimized for the
range from 100 to 300 Hz to 1 to 3
kHz, with ported LF systems handling Call or e-mail NOW to find out just how fast your tests can be!
the range below. Rapid flare HF horns
are now employed, and their distortion
characteristics, level for level, may be Prism Media Products Limited Prism Media Products Inc.
William James House, 21 Pine Street,
up to 10 dB lower than the older hard- Cowley Road, Cambridge. Rockaway, NJ.
CB4 0WX. UK. 07866. USA.
ware [4]. New digital methods of signal
control have made multiway systems Tel: +44 (0)1223 424988 Tel: 1-973 983 9577
Fax: +44 (0)1223 425023 Fax: 1-973 983 9588
more acceptable than they were in the
sales@prismsound.com www.prismsound.com
days of passive dividing networks,
through time alignment and steep
crossover slopes.
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April 415
 HISTORICAL PERSPECTIVES AND TECHNOLOGY OVERVIEW
OF LOUDSPEAKERS FOR SOUND REINFORCEMENT

Fig. 8. The cone driver. Section view (A); equivalent circuit (B); radiation impedance of a cone in a large wall (C); power
response of system (D).

sients and other telegraphic signals. He
remarked at the time that it could be
used “for moving visible and audible
signals.”
Half a century later in 1925, Rice
and Kellogg of General Electric
described “a new hornless loud-
speaker” that resembled that of
Siemens—a similarity that prompted
Rice to say: “The ancients have stolen
our inventions!” [37].
The key difference in the Rice and
Kellogg design was the adjustment of
mechanical parameters so that the fun-
damental resonance of the moving sys-
tem took place at a lower frequency
than that at which the cone’s radiation
impedance had become uniform. Over
this range, the motion of the cone was
416
mass controlled, and the cone looked with frequency to approximately ka = 2,
into a rising radiation impedance. This above which point it is essentially con-
in effect provided a significant fre- stant (ka is equal to cone circumference
quency region of flat power response divided by wavelength, or, 2πa/λ). Sys-
for the design. Details of this are shown tem response is shown in Fig. 8D, and
in Fig. 8. system efficiency over the so-called pis-
ton band is given [49] as:
Region of Flat Power Response
Fig. 8A shows a section view of the
cone loudspeaker with all electrical,
mechanical, and acoustical parame-
ters labeled. The equivalent circuit is
shown in Fig. 8B; here the mechani-
cal and acoustical parameters are
shown in the mobility analogy.
When mounted in a large baffle,
the moving system looks into a com-
plex acoustical load as shown in Fig. Fig. 9. Illustration of mutual coupling of LF
8C. The resistive component rises drivers.
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April
 HISTORICAL PERSPECTIVES AND TECHNOLOGY OVERVIEW
OF LOUDSPEAKERS FOR SOUND REINFORCEMENT
η = [ρ0(Bl)2SD2/RE]/2πcMMS2
(1)

where ρ 0 is the density of air

(kg/m3), (Bl)2/RE is the electromechani-
cal coupling coefficient (N2/W), c is the
speed of sound (m/s), SD2 is the area of
the cone, and MMS is the mass of the
moving system (kg).
The larger the coupling coefficient,
the lower the resonant Q at f0 and the
higher the piston band efficiency will
be. Likewise, the higher the Q at f0, the
lower the piston band efficiency.
Depending on the application, both
kinds of response may be useful to the
design engineer.
It can easily be seen that, for maxi-
mum extension of the piston band, the
lower f0 must be, and the lower the sys-
tem efficiency will be. The efficiency-
bandwidth product, for a given cone
diameter and coupling coefficient, thus
tends to be constant over a relatively
large range.

Mutual Coupling Fig. 10. The ported system. Section view (A); equivalent circuit (B); port and cone
In the LF range over which their contributions to total output (C).
response is essentially omnidirectional
(ka = 0.2 or lower), a doubling of
closely spaced driving units will result
in an increase in acoustical output of 3
dB for a fixed input power reference
level [39, 48, 52, 53]. The progression
in efficiency increase is shown in Fig. 9
for one, two, and four LF transducers,
respectively. In each case, the electrical
power delivered to each ensemble of
drivers is constant. Assume that the ref-
erence power fed to the single driver is
one watt; then for the set of two
drivers, the power per driver is 0.5
watt, and for the set of four, the power
per driver is 0.25 watt. Fig. 11. Power compression in a 380-mm-diameter LF driver. Curves for 1 and 100
One may imagine that, in the two- watts input are superimposed and displaced by 20 dB. (Data courtesy JBL.)
driver case with both drivers wired in
parallel, those two drivers have, in a give something for nothing, but there tion of power doubling for each two-
sense, coalesced into a new driver— are clear limits to its effectiveness. times increase in drivers is accurate
one with twice the cone area, twice the With each doubling of cone area, the only at very low frequencies and only
moving mass, and half the value of RE. ka = 0.2 upper response frequency if the efficiency values are low to
Thus, by Equation 1, the efficiency will corner moves downward approxi- begin with.
have doubled. For the case where the mately by a factor of 0.7, since this is
two drivers are wired in series, the the reciprocal of the value by which Distortion
analysis goes as follows: The new the effective cone radius has
driver has twice the cone area, twice increased. As the process of adding Mechanical Effects
the moving mass, four times the (Bl)2 drivers is continued, in the limit it can The primary distortion mechanism in
product, and twice the value of R E. be shown that the efficiency of an cone transducers is due to mechanical
Again, by Equation 1, there will be a ensemble of direct radiators cannot stress–strain limits. Small identified a
doubling of efficiency. exceed a value of 25% [38]. Because practical mechanical displacement
Mutual coupling often appears to of these constraints, the approxima- limit from rest position in the axial ➥
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April 417
 HISTORICAL PERSPECTIVES AND TECHNOLOGY OVERVIEW
OF LOUDSPEAKERS FOR SOUND REINFORCEMENT
Fig. 12. Power compression. Output level versus time for three LF driver
designs. (Data courtesy JBL.)
direction as the excursion at which
10% harmonic distortion is reached.
This limit is known as xMAX. While a
loudspeaker may be operated beyond
this displacement limit, at least on a
momentary basis, the 10% linearity
departure is generally recognized as a
safe limit for good engineering prac-
tice. Since cone displacement tends to
increase as the inverse square of
frequency down to the f0 region, it is
easy to see how the x MAX limitation
may easily be encountered in normal
operation.
The onset of the cone displacement
limit at low frequencies can be allevi-
ated by using ported LF enclosures. The
nature of this design is shown in Fig.
10. A section view of a ported system is
shown in Fig. 10A, and the equivalent
circuit is shown in Fig. 10B. The design
relies on controlling the Helmholtz res-
onance of the enclosure to provide an
“assisted output,” via the port, that min-
imizes cone motion (and thus distor-
tion) at low frequencies, as shown in
Fig. 10C. Thiele–Small parameters are
universally used today to synthesize
these systems.
Virtually all commercial PC design
programs for ported systems will indi-
cate transducer displacement limits so
that the design engineer will always be
aware of whether a system, while still
on the drawing board, will go into dis-
placement overload before it reaches its
thermal limit. Good engineering prac-
tice demands that a ported system
remain thermally limited down to f0.
418
Fig. 13. Demagnetization plots for three
loudspeaker magnetic materials. Alnico V (1);
ferrite (2); neodymium-iron-boron type (3).
Below that frequency the electrical
drive signal is generally rolled off to
avoid subsonic over-excursions of the
cone.
A secondary mechanical distortion
effect will be seen when the voice coil
driver is far enough out of the gap so
that there is a momentary loss of Bl
product at peak excursion values. The
effect is asymmetrical and gives rise
to both even and odd distortion
components.
Port Turbulence in Vented Systems
In vented systems the ultimate output at
low frequencies may be limited not by
considerations of maximum cone
excursion, but rather by air turbulence
in the enclosure port when the system
is operating at the tuning frequency
[34]. A tentative limit here is to restrict
the port air particle velocity so that its
peak value does not exceed about 5%
of the speed of sound. In general, ports
should be designed with contoured
boundaries to minimize turbulence and
the noise and losses it often produces.
Significant studies of port turbulence
and its minimization through tapering
the port tube’s cross-section area have
been carried out by Vanderkooy [50,
51], Salvatti and Button [45], and
Roozen et al. [44].
Thermal Effects
Modern cone transducers intended for
heavy-duty professional applications
take advantage of newer materials and
adhesives to make them more immune
to thermal failure. Thermal failure is
reached when the power dissipated in
the voice coil as heat cannot be
removed at a sufficient rate to maintain
a safe operating temperature. A great
deal of loudspeaker development has
gone into designing structures and
moving elements that are not only
resistant to heat but aid in its removal
from the transducer [32, 36].
For most applications in sound rein-
forcement, the effects of loudspeaker
heating are more likely to result in
component failure than those associ-
ated with displacement limitations.
Dynamic linearity or power compres-
sion are terms used to describe the
effects of heating on audio performance
[34]. The data shown in Fig. 11 pre-
sents the frequency response of a single
380-mm LF transducer with inputs of 1
watt and 100 watts. In each case the
chart recording of the levels has been
adjusted to account for the 20-dB offset
between the curves. In this manner the
response differences can be clearly
seen. If there were no dynamic com-
pression, the two curves would lie one
on top of the other. As it is, the pro-
gressive heating results in an increased
value of R E, which lowers the effi-
ciency. In extreme cases, the increase
in RE can result in changes in the LF
alignment, which may be clearly audi-
ble as such.
Another way of viewing power com-
pression is shown in Fig. 12. Here, sev-
eral 380-mm transducers have been
driven with a wide-band signal, and the
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April
 HISTORICAL PERSPECTIVES AND TECHNOLOGY OVERVIEW
OF LOUDSPEAKERS FOR SOUND REINFORCEMENT
Fig. 15. The Altec-Lansing Duocone. Section view of cone (A);
equivalent circuit (B).
tion performance of a
transducer [33]. The
primary effect is the
variation in the mag-
netic operating point
that comes as a result
Fig. 14. Performance with same moving system in three
of signal current in
different motor assemblies. Alnico V (A); traditional ferrite the voice coil as it
(B); ferrite with undercut polepiece and aluminum modulates the mag-
shorting ring (C). (Data courtesy JBL.) netic structure’s oper-
ating point. This hap-
effect of temperature rise has been plot- pens to some degree in all designs,
ted against time. For each transducer but the effect on acoustical perfor-
design, the reduction in output level mance depends on the degree of flux
eventually reaches an asymptotic value modulation of the static magnetic
that depends on how effectively heat field. The general result of flux modu-
can be removed from the voice coil and lation is to increase the level of sin-
magnet structure. In general, larger gle-ended distortion, which is largely
diameter voice coils remove heat more second harmonic.
efficiently than smaller ones; addi- Fig. 13 shows the demagnetization
tional measures, such as increasing curves for three magnetic materials,
convection cooling and increasing Alnico V, ferrite, and neodymium-
radiation to outside mechanical parts based systems. With the Alnico V sys-
of the driver, can be helpful as well. tem, the flux curve has a moderate
As the temperature of a ferrite mag- slope in the operating region, which is
netic structure increases, both flux den- near the intersection with the B-axis; at
sity and efficiency are reduced. The that position there is little tendency for
effect is similar to the resistance flux modulation to be a problem. How-
increase caused by voice coil heating. ever, a strong input signal can result in
If the temperature rise in the ferrite permanent demagnetization if the oper-
material has been moderate, normal ating point is forced downward on the
performance may be restored upon steep portion of the curve to the left of
cooling. the normal operating point.
With the ferrite magnet there is a cer-
Distortion in Transducer Magnetic tain amount of flux modulation due to
Systems the uniform slope of the curve. With
Many aspects of the direct radiator’s the high-energy neodymium materials,
magnetic system can affect the distor- the demagnetization curve is normally
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April
located high enough in the second
quadrant of the B-H graph that the
magnetic circuit is very likely to be
operating at or near saturation. In this
case the degree of flux modulation will
be minimal.
Fig. 14 shows how a typical problem
was solved. The distortion data shown in
Fig. 14A is that of an older 300-mm-
diameter LF transducer operating with
an Alnico V magnet structure. Keeping
the same moving system but changing to
a typical ferrite magnet structure yields
the data shown in Fig. 14B. Note the
significant increase in mid-frequency
distortion.
The data in Fig. 14C shows the
effect of a ferrite magnet system out-
fitted with undercut polepiece geome-
try and a large, low-resistance alu-
minum flux shorting ring placed at the
base of the polepiece. The significant
reduction in second-harmonic distor-
tion results from the setting up of an
induction current in the shorting ring
that counteracts the normal tendency
of voice coil current to shift the mag-
netic operating point.
Other magnetic distortion effects
include: the generation of eddy cur-
rents in local iron structures, which
results in an increase in third-har-
monic distortion and I2R losses; and
inductance modulation of the voice
coil, due the varying amount iron
instantaneously surrounded by the
voice coil, which results in increased
second-harmonic distortion. ➥
419
 HISTORICAL PERSPECTIVES AND TECHNOLOGY OVERVIEW
OF LOUDSPEAKERS FOR SOUND REINFORCEMENT
Fig. 16. Polar plots for a piston mounted in a wall. (Data after Beranek, 1954.)
Thermodynamic and FM Distortion
Effects
Thermodynamic distortion, or air over-
load, is present in relatively small
quantities in direct radiator systems and
may be disregarded in the normal oper-
ation of cones and domes.
Frequency modulation (FM) com-
ponents are more likely to occur,
especially at low to mid frequencies,
where high cone excursions at lower
frequencies may modulate higher fre-
quencies as the cone’s velocity attains
a significant percentage of the speed
of sound. The effect is quite notice-
able in single-cone systems but is
minimized in multiway systems.
Beers and Belar [30] and Klipsch [40]
were among the first to describe this
phenomenon in detail.
Another aspect of thermodynamic
distortion present in sealed direct radia-
tor systems is due to the nonlinearity of
the enclosed air spring. In general, if
the maximum instantaneous change in
an enclosed volume can be limited to
about 0.5%, the effect of the nonlinear-
ity can be ignored [34]. The type and
amount of enclosure damping material
in a sealed enclosure has the additional
effect of increasing the actual enclosed
volume. The work of Leach [41] is sig-
nificant in this area.

The Decoupled Cone

Over the years loudspeaker designers
have observed that, at high frequen-
cies, the cone ceases to move as a sin-
gle unit, but rather breaks up into more
complex motions. These result in an
effective lighter moving mass at high
frequencies, extending the HF
response of the system. This was first
commercialized by the Altec Duocone
loudspeaker [28]. Fig. 15A shows a
section view of the cone profile used in
the Duocone loudspeaker, and a mobil-
ity mechanical circuit is shown in Fig.
15B. The modern soft-dome HF unit
exploits this effect more predictably
through high damping of the moving
system. Finite element analysis (FEA)
provides a means of analyzing in detail
the directionality of loudspeaker cones
and domes, taking into account multi-
ple breakup effects. Many low-cost
drivers intended for ceiling installation
in distributed systems make good use
Fig. 17. Polar plots for a piston mounted at the end of a long tube. (Data after of decoupling for attaining wider cov-
Beranek, 1954.) erage at HF.
420 J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April
 HISTORICAL PERSPECTIVES AND TECHNOLOGY OVERVIEW
OF LOUDSPEAKERS FOR SOUND REINFORCEMENT
In recent years, the notion of decou-
pling has resurfaced with the dis-
tributed mode loudspeaker (DML).
Here, a panel of light, stiff, and fairly
lossless material is driven by a mass-
loaded transducer. When operated in
the frequency range over which the
panel exhibits a dense array of higher-
order two-dimensional bending modes,
the resulting power output and disper-
sion are uniform over a frequency
range defined by:

Upper frequency limit, fmax = RP/2πMc

(2A)

Lower frequency limit, fmin = RP/2πMm

(2B)
where RP is the radiation impedance of
the panel and Mc and Mm are, respec-
tively, the masses of the driven system
and the magnetic actuator.
Due to its random, diffuse nature, the
response of a typical DML panel
departs from that of a typical dynamic Fig. 18. The Altec Extenda-Voice gradient loudspeaker system. Section view (A);
physical circuit (B); nominal off-axis response curves (C).
driver in several regards: the modulus
of electrical impedance is quite uniform
over the normal system passband; the
solid-angle radiation pattern is quite
uniform over the normal system pass-
band; the impulse response for a typical
panel is fairly long, upwards of 25
msec, however it is free of ringing at
any specific frequency; commensurate
with the long impulse train, the ampli-
tude response will exhibit, in all direc-
tions of radiation, many fine response
dips on the order of 6 dB [29].
With its unique combination of
acoustical properties and relatively low
cost, the DML offers great promise in
the area of distributed system design.

Directional Properties of Direct

Radiators
Assuming that the moving systems
have only a single degree of freedom,
the polar data in Fig. 16 shows the
theoretical directionality of a piston
mounted in a large baffle as a func-
tion of ka. The on-axis directivity
index (DI) of each directional pattern
is also shown in the figure [1, 42].
Similar data is shown in Fig. 17 for a
piston mounted at the end of a long
tube. These two conditions simulate
normal 2π and 4π mounting condi-
tions for loudspeakers. For many rou-
tine system design applications, this
data is sufficiently accurate.
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April
Fig. 19. Radiation resistance for conical, exponential, and hyperbolic horns.
Gradient Loudspeakers
The simplest gradient loudspeaker is an
unbaffled cone transducer operating as
a dipole. The natural directional
response for a dipole is a cosine pattern,
or a “figure-eight” [43]. Dipoles are
very inefficient and have not been
widely used in sound reinforcement
applications; however, there are certain
applications where their specific radia-
tion pattern could be very useful. Dur-
ing the early 1970s, Altec produced a
modified gradient loudspeaker in which
one side of the dipole operated in series
with an acoustical delay path. The delay
produced a hypercardioid pattern in
much the same manner as a typical
dynamic microphone does.
Fig. 18A shows a section view
through the hypercardioid loudspeaker.
The output from the rear of the trans-
ducer is sent through a constant delay,
provided by the path length and the
resistive elements in that path. The result
of the delay is that, at 135°, output from
front and back of the transducer will ➥
421
 HISTORICAL PERSPECTIVES AND TECHNOLOGY OVERVIEW
OF LOUDSPEAKERS FOR SOUND REINFORCEMENT
Fig. 20. Real and imaginary components of horn impedance for a long exponential
horn (A) and a short exponential horn (B).
Fig. 21. Plane wave tube measurements of a compression driver showing
amplitude response and impedance.
effectively cancel. For useful output
from the front of the transducer, the sig-
nal fed to the system must be equalized
with a 6-dB-per-octave rise for each
halving of frequency to compensate for
the diminishing gradient. The equivalent
physical circuit of the loudspeaker is
shown in Fig. 18B, and off-axis polar
data is shown in Fig. 18C. A system
such as this would normally be used for
speech purposes in highly reverberant
spaces where the loudspeaker’s DI of 6
dB at LF would work to its advantage.
Vertical stacks of the device can
increase total output capability as well as
increase the on-axis DI.
HORNS AND COMPRESSION
DRIVERS
Early Development
Many engineers and physicists have
contributed to horn and compression
driver development over the years.
Early versions of the horn were used by
many tinkerers who basically did not
understand how the horn worked—they
knew only that somehow the horn
422
increased acoustical output [58]. The
first example of thorough engineering
was carried out by Bell Telephone Lab-
oratories [83], working from the model
of horn impedance described by Web-
ster [81]. Significant later development
was carried out by Klipsch [66], who
designed a remarkably compact bass
horn, and Salmon [76, 77], who
described the impedance characteristics
of several important horn flares, includ-
ing the hyperbolic, or Hypex, profile
[68]. Geddes [55] sought to position
Webster’s model as a special case
within a broader context.
Fig. 19 shows the real part of the
radiation impedance for hyperbolic,
exponential, and conical horn profiles.
Here, only the exponential and hyper-
bolic profiles provide useful output at
low frequencies. In our discussion we
will restrict ourselves to the exponen-
tial profile, since it has found almost
universal application over the years.
Fig. 20A shows the real and imagi-
nary parts of throat impedance for a
long exponential horn. For a horn of
practical length, we might observe
impedance components such as those
shown in Fig. 20B. The slight peaks
and dips in response are due to reflec-
tions from the mouth of the horn back
to the throat. There is an optimum
mouth size for a horn of specific cutoff
frequency to minimize reflected waves
from the horn’s mouth [63].
Theoretical Modeling
The compression driver is designed to
match the impedance of the electrome-
chanical system to the throat of the horn,
and the radiation impedance, reflected to
the electrical side of the circuit, is:
RET = [ST(Bl)2]/ρ0cSD2 (3)
where ST is the area of the driver throat
and S D is the area of the driver
diaphragm. The phasing plug in the
driver is the means by which the ratio
of the two areas is adjusted for best HF
response.
When the driver is attached to the
horn, the efficiency in the range where
the horn’s radiation impedance is
essentially resistive is:
η = (2RERET)/(RE + RET)2 (4)
where RE is the voice coil resistance.
When the voice coil resistance is made
equal to the radiation resistance, the
efficiency of the driver over its normal
pass-band will in theory be 50%. In
practice, efficiencies of the order of
30% can be achieved in the mid-
range—and this is only about
2 dB below the theoretical maximum.
Region of Flat Power Output
The data of Fig. 21 shows the normal
power response for a compression
driver/horn combination when the
horn’s throat impedance is resistive.
The LF limit is due to the primary reso-
nance of the driver; for a typical HF
compression driver this may be in the
range of 500 Hz.
The principal midband rolloff com-
mences at what is called the mass break
point, fHM, given by:
fHM = (Bl)2/πREMMS (5)
where MMS is the mass of the moving
system. For most HF compression
drivers the mass breakpoint takes place
in the range of 2500 to 4500 Hz. It is
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April
 HISTORICAL PERSPECTIVES AND TECHNOLOGY OVERVIEW
OF LOUDSPEAKERS FOR SOUND REINFORCEMENT
considered a fundamental limit in HF
drivers, inasmuch as today’s magnetic
flux densities are normally maximized
in the range of 2 tesla, and low-moving
mass is limited by metals such as tita-
nium and beryllium that are not likely
to be replaced in the near future.
Two additional inflection points are
often seen in the HF driver response
curve: one is due to the volume of the
front air chamber in the driver, the
space between the diaphragm and the
phasing plug. Its effect on response
may be seen as low as 8 kHz in some
drivers. Voice coil inductance may
cause an additional HF rolloff at high
frequencies. This may be compensated
for through the use of a silver or cop-
per shorting ring plated on the pole-
piece in the region of the voice coil. Fig. 22. Second-harmonic distortion in two horn systems, using a compressed
fundamental. (Data courtesy JBL.)
(See Distortion in Transducer Mag-
netic Systems, page 419.)

Cone-Driven Horns
From the earliest days, cone transduc-
ers have been employed as horn drivers
[71, 72]. The theoretical principles that
govern the design parameters for horn
drivers apply equally to the adaptation
of cone drivers as well as to purpose-
designed compression drivers. Keele
[60] presented a straightforward and
useful analysis of LF horn design using
both Thiele–Small and traditional elec-
tromechanical parameters. Leach [69]
summarized Keele’s work, together
with Small’s approach to the subject
[78], and addressed other factors such
as reactance annulling.
Fig. 23. Plane wave tube amplitude response of three compression drivers.
Reactance Annulling
In some compression driver designs, a Distortion sound velocity at elevated temperatures
mechanical stiffness in the form of a The dominant cause of distortion in under adiabatic conditions.
small air chamber is located behind the compression driver-horn systems is due Thuras, Jenkins, and O’Neil [80] and
driver’s diaphragm. The mechanical to thermodynamic, or air, overload Goldstein and McLachlan [57] ana-
reactance resulting from the stiffness [75]. This comes as a result of lyzed the problem, leading to a simpli-
cancels in part the mass reactance por- extremely high pressures that exists at fied equation that gives the percent
tion of the radiation impedance, result- the horn throat: second harmonic distortion in horn
ing in a more resistive impedance in the systems:
region of the cutoff frequency. The LP = 94 + 20 log (WA(ρ0c)/ST)0.5 (6)
effect of this is greater acoustic output % 2nd HD = 1.73(f/fC) √IT x 10-2
in the horn cut-off frequency range for where WA is the acoustical power gener- (7)
a given drive signal [73, 74]. ated and ST is the throat area (m2).
Reactance annulling is not normally For example, in plane wave propaga- where IT is the intensity in watts per
used in HF compression drivers, but it tion, an intensity of one watt per square square meter at the horn’s throat, f is
is used in the design of bass horns, most centimeter will produce a sound pres- the driving frequency, and fC is the cut-
notably in the case of the Klipschorn, sure level of 160 dB LP. For levels in off frequency of the horn.
where the normal response associated this range, successive pressure peaks Fig. 22 presents measurements of the
with a 47-Hz flare rate is extended are tilted forward as they propagate second-harmonic distortion produced
down to about 40 Hz [66]. down the horn due to the increase in by two horns of differing flare rates. ➥
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April 423
 HISTORICAL PERSPECTIVES AND TECHNOLOGY OVERVIEW
OF LOUDSPEAKERS FOR SOUND REINFORCEMENT
kind of distortion can be defined mathe-
matically, a model can be implemented
and used to predistort the signal, result-
ing in reduced distortion in the system’s
output over a given power operating
range. Klippel [65] describes some of
the techniques for accomplishing this.

The Role of Secondary

Fig. 24. Directivity of various exponential horns. (Data from Olson, 1957.)
The fundamental output in each case
was held constant at a level of 107 dB
at a distance of one meter, and the sec-
ond-harmonic distortion has been
raised 20 dB for ease in reading. The
scale on the right ordinate indicates the
second-harmonic distortion in percent-
age. The cutoff frequency of the horn
used for the data shown in Fig. 22A is
70 Hz, while that of the horn used for
the data shown in Fig. 22B is approxi-
mately 560 Hz. The average difference
in distortion is 8 dB.
A horn with a high cutoff frequency
has a rapid flare rate and as such may
lack good directional control at low fre-
quencies. This is a tradeoff that the
design engineer has to reconcile for a
variety of applications. For example,
sound-reinforcement applications
require specific pattern control in the
range from 300 Hz upward, while
music-monitoring applications may
require horn pattern control no lower
than about 800 Hz.
If the exact mechanism for a given
Resonances
As shown earlier in Region of Flat
Power Repsonse (page 416), the power
response of a horn driver is flat up to its
mass break point, above which the
response rolls off 6 dB per octave. How-
ever, beneficial secondary resonances
may be used to increase the driver’s out-
put above this point [64, 70]. These res-
onances generally occur in the surround
sections of the diaphragm and are
decoupled from the diaphragm itself. As
in the case of decoupled resonances in
cones discussed earlier, the lowering of
moving mass at higher frequencies can
result in a considerable increase in use-
ful HF response. Fig. 23 shows the
response for three compression drivers,
all with 100-mm-diameter diaphragms
and mounted on the same horn. Driver
A has an aluminum diaphragm and half-
roll surround. The secondary resonance
is about 9 kHz. Response is maintained
fairly flat to that frequency, falling off
rapidly above. Driver B has a beryllium

Fig. 25. Beamwidth and directivity data for typical radial horn (A) and slant-plate lens system (B).

424 J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April

 HISTORICAL PERSPECTIVES AND TECHNOLOGY OVERVIEW
OF LOUDSPEAKERS FOR SOUND REINFORCEMENT

Fig. 26. Directivity of a uniform coverage horn. Basic

directivity regimes (A); beamwidth and directivity for a 90° Fig. 27. Attenuation of sound with distance from point, line, and
by 40° uniform directivity horn (B). planar sound sources.

diaphragm with a half-roll surround.
Note that, due to the greater stiffness and
lower mass of the material, the sec-
ondary resonance has shifted out to
about 17 kHz. Driver C has an alu-
minum diaphragm with distributed sur-
round geometry that moves the sec-
ondary resonance to beyond 20 kHz,
resulting in smooth, extended response
within the normal audio band with no
pronounced peaks [70].
Directional Response
The basic exponential horn exhibits
directional response as shown in
Fig. 24. From the earliest days it was
recognized that directional characteris-
tics were a key element of loudspeaker
performance [84]. Over decades of
development, numerous methods have
been used to improve directional per-
formance at high frequencies for sound
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April
reinforcement applications. For multi-
cellular horns, in the early days [83],
groups of exponential cells, each about
15° wide in each plane, were clustered
together to define a specific solid radi-
ation angle; this produced excellent
results at mid-frequencies, but there
was pronounced “fingering” of the
response along the cell boundaries at
higher frequencies. For radial horns, in
this application, the horn’s horizontal
profile is conical, with straight, radial
sides defining a target coverage angle
and the vertical profile is tailored to
make a net exponential profile along
the horn’s primary axis; the nominal
horizontal and vertical -6 dB
beamwidth of a radial horn is shown in
Fig. 25A [71]. For acoustical lenses, a
slant-plate acoustical lens can be
placed at the mouth of an exponential
horn to diverge the exiting waves in
one dimension, as shown in Fig. 25B
[54, 67].
Constant Directivity Horns
Also known as uniform coverage or
constant coverage horns, these designs
date from the mid-1970s to the early
1980s [59, 62, 79]. The basic design
common to a number of manufacturers
uses a combination of exponential or
conical throat loading, diffraction wave
guide principles, and flared termina-
tions to produce uniform nominal cov-
erage angles in the horizontal and verti-
cal planes. The general shape of the
beamwidth curve is shown in Fig. 26A,
as it applies to the horizontal and verti-
cal planes independently. Fig. 26B
shows the measured beamwidth and DI
of a typical constant directivity horn
with nominal 90°-by-40° pattern con-
trol. Within certain limitations, ➥
425
 HISTORICAL PERSPECTIVES AND TECHNOLOGY OVERVIEW
OF LOUDSPEAKERS FOR SOUND REINFORCEMENT

Fig. 28. Directivity of a four-element vertical line array with 0.2 meter separation between driver centers. 200 Hz (A); 350 Hz
(B); 500 Hz (C); 1 kHz (D); directivity factor for arrays of 4, 6, 8, and 10 elements (E).

acoustic waveguide theory has pro-
posed an alternate approach to achiev-
ing similar goals [55, 56].
ARRAYS
Both horns and direct radiators may
be treated the same in terms of array-
ing. In this section we will examine
some useful concepts. Single-ele-
ment, line, and planar arrays differ in
their radiation characteristics over
distance, as shown in Fig. 27. At long
wavelengths, the simple inverse
square relationship of a point source
A is modified by a line array as
shown at B, and a very large planar
array will show little attenuation with
distance up to limits proportional to
the array dimensions [98]. A finite
planar array will have the characteris-
tics shown at D. Long horizontal line
arrays have been placed above
prosceniums in performance spaces
to extend the range of a high direct-
to-reverberant ratio toward the rear of
the space; large planar arrays are the
mainstay of mega-event music rein-
forcement [87].
The Simple Line Array
Kuttruff [95] describes the polar
response of a line array of omnidirec-
tional sources in the plane of the array as:
R (θ) = (sin [1/2 Nkd sin θ])/(N sin
[1/2 kd sin θ]) (8)
where N is the number of elements in
the array, k is 2πf/c, d is the spacing of
the elements in the array, c is the speed
of sound, and θ is the measurement
angle in radians. For four elements as
shown in Fig. 28, the polar response ➥

C
A B

Fig. 29. Tapering of arrays. Electrical tapering (A); acoustical tapering (B); tapering through component rotation (C).

426 J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April

 HISTORICAL PERSPECTIVES AND TECHNOLOGY OVERVIEW
OF LOUDSPEAKERS FOR SOUND REINFORCEMENT
Fig. 30. The Bessel array. Layout (A);
series wiring diagram (B); parallel wiring
diagram (C); polar response (D). (Data
courtesy Keele, 1990.)
is shown in A through D. The directiv-
ity factor is shown at E.
The four-element array will exhibit
good pattern control over the range
from d/λ 0.5 to 2.0. At higher frequen-
cies the pattern will exhibit narrowing
and lobing, and simple arrays of more
than six elements will generally have
unsatisfactory characteristics.
Tapering the Line Array
The accepted method of extending the
uniform coverage range of a line array
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April
Fig. 31. Grateful Dead “Wall of Sound” direct radiator system, 1974, using line,
planar, and arc segment arrays. Individual discrete systems were employed for
each instrument separately from the vocal reinforcement. (Photo courtesy
Richard Pechner.)
is through frequency tapering, or
shaping, to allow the array to, in
effect, reduce in size with rising fre-
quency. Some techniques are shown
in Fig. 29. Electrical frequency
tapering is shown at A, and acousti-
cal frequency tapering is shown at
B [94]. A unique “barber-pole”
array is shown at C [93].
The Product Theorem
The product theorem [91] states that
an array composed of uniform direc-
tional elements will exhibit the same
far-field response as a like array of
omnidirectional elements multiplied
by the directional properties of one
of the single directional elements.
This is another way of stating the
principle of superposition, and is
useful in estimating the directional
response of complex arrays.
The Bessel Array
Franssen [88] describes an array of
elements whose amplitude drive char-
acteristics are derived from Bessel
coefficients [90, 92]. A simple 5-ele-
ment array is shown in Fig. 30A, and
simplified wiring diagrams to derive
the drive coefficients are shown at B
and C. The far-field response of the
array, modeled with omnidirectional
sources, is shown at D. Note that the
response is essentially omnidirectional
over a 100-to-1 wavelength ratio.
Via the product theorem, each omni-
directional element could be replaced
with a directional element, all oriented
in the same direction, with the resulting
response of the array exhibiting the
chosen directional characteristic over
the same 100-to-1 wavelength ratio.
The Bessel array has potential for
speech reinforcement in live spaces. Its
phase response varies with angle and
frequency, however, and thus it may be
difficult to integrate the concept into a
system that includes standard radiating
elements.
Very Large Arrays for Music
Concert sound reinforcement in very
large venues, indoors or out, requires
large arrays, and the accepted method
of assembling these arrays is to use
building blocks that are each relatively
full-range units. Thus, the assembled
system, normally resembling a large
plane, or sets of planes with curved sec-
tions connecting them, has much in
common with the principles described
in Fig. 31 [87]. As an example of this
we show in Fig. 32 an elevation view of
a typical large vertical array (A) along
with the on- and off-axis response (B)
measured in an arc along the listening
plane [89].
The great virtue of these systems is
their ability to deliver very high ➥
427
 HISTORICAL PERSPECTIVES AND TECHNOLOGY OVERVIEW
OF LOUDSPEAKERS FOR SOUND REINFORCEMENT

B
A

Fig. 32. A large array for music reinforcement. Physical layout (A); off-axis response on the ground plane (B). (Data courtesy
Gander and Eargle.)

Fig. 33. Performance of a continuous array 3

meters in length. Attenuation with distance at Fig. 34. Performance of a spiral array. Side view of array (A); polar response of
10 kHz (A); polar response at 10 kHz (B). array at 500 Hz, 1, 2, and 4 kHz (B). (Data courtesy M. Ureda.)
428 J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April
 HISTORICAL PERSPECTIVES AND TECHNOLOGY OVERVIEW
OF LOUDSPEAKERS FOR SOUND REINFORCEMENT
A B
35. A programmable array. View of array (A); signal flow diagram (B); examples of variable polar response (C). (Data courtesy
Eastern Acoustics Works.)
sound pressure levels at considerable
distances with relatively low distortion.
The primary defect is the dense comb
filtering (lobing) and “time smearing”
that inevitably results from such a multi-
plicity of sources covering a given lis-
tening position. Actually, since the
required acoustic power cannot be
achieved by a single source, the aim
here should be to keep the coverage at
each listening position as uniform as
possible. The greater the number of
effective sound sources, the finer the
lobing patterns and the more uniform
the received signal will be. Ideally, we
would like the interference peaks and
dips among the elements to be well
within the ear’s critical bandwidths.
Continuous Line Arrays
A continuous line array is an approxi-
mation of a uniformly illuminated “rib-
bon” of sound, and its directional
behavior in the far field can be deter-
mined by equation (8). At low frequen-
cies, the far-field for a straight line
array begins approximately at a dis-
tance equal to the array length divided
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April
C
by π [98]. For progressively shorter
wavelengths, this distance increases
according to the following equation:
r = l2f/700 meters (9)
where l is the array length, and r and l
are in meters [99].
Fig. 33A shows the attenuation pat-
tern with distance from a straight
3-meter array at 10 kHz, and the polar
response in the far field of that array is
shown at B. Note that the beamwidth (-
6-dB) is 0.8°, as given by the equation:
θline array = 2 sin-1(0.6λ/l) (10)
where l is the length of the array and λ
is the wavelength [99].
The pronounced HF beaming of
straight arrays is of course a liability,
and articulation of the array is one way
of dealing with the problem
J and Spiral Arrays
Ureda [100] describes a J array as hav-
ing two distinct portions: a straight
upper section and a uniformly curved
lower section. Each segment of the
array will act independently, with the
upper section producing a highly direc-
tive HF beam for distant coverage and
the lower section producing broader
radiation for near coverage.
Ureda proposes the spiral array for
more uniform overall coverage. The
spiral array is continuously curved
from the top down, beginning with
small angular increments, which
increase downward in arithmetic fash-
ion. Fig. 34A shows a side view of
such an array with a total length of 5
meters and a terminal angle of 45°.
The directivity function is remark-
ably uniform with frequency over about
a decade. Fig. 34B shows a group of
polar plots from 500 Hz to 5 kHz.
Steerable Arrays
The very large arrays for music dis-
cussed earlier consist of elements
equally driven in terms of level and
bandwidth, and the directional proper-
ties are due entirely to the spatial rela-
tionships among elements. A steerable
array is one in which the elements are
fixed in space, with relative drive ➥
429
 HISTORICAL PERSPECTIVES AND TECHNOLOGY OVERVIEW
OF LOUDSPEAKERS FOR SOUND REINFORCEMENT
levels, signal delay, and frequency
tapering individually adjustable for
each transducer.
Relatively simple arrays can be
reconfigured, through sequential tim-
ing, to steer their beams as needed [85,
86, 96, 97]. While far-field modeling
may be fairly simple, the fact that many
listeners are seated in the transition
region between near and far fields
makes the problems of reconfiguration
and uniformity of coverage fairly com-
plex to estimate.
The relatively small system shown in
Fig. 35 can be configured via a PC by
the user as required. The system profile
and signal flow diagram are shown at A
and B, and a family of typical far-field
polar plots is shown at C. Systems such
as these, large or small, are presently
used to solve intelligibility problems in
a variety of large reverberant spaces.
REFERENCES AND
SUPPLEMENTAL
BIBLIOGRAPHY
Historical Perspectives
[1] Beranek, L. Acoustics, John
Wiley & Sons, New York, 1954. Cor-
rected edition published by the Ameri-
can Institute of Physics for the Acous-
tical Society of America, 1986.
[2] Beranek, L., “Loudspeakers and
Microphones,” J. Acoust. Soc. Am.,
26:5 (1954).
[3] Clark, L. and Hilliard, J., “Head-
phones and Loudspeakers,” Chapter
VII in Motion Picture Sound Engi-
neering, D. Van Nostrand, New York,
1938.
[4] Eargle, J. and Gelow, W., “Per-
formance of Horn Systems: Low-Fre-
quency Cut-off, Pattern Control, and
Distortion Trade-offs,” presented at
the 101st AES Convention, Los Ange-
les, 8-11 November 1996; preprint
4330.
[5] Engebretson, M. and Eargle, J.,
“Cinema Sound Reproduction Sys-
tems: Technology Advances and Sys-
tem Design Considerations,” SMPTE,
91:11 (1982). Also see AES preprint
1799.
[6] Engebretson, M., “Low Fre-
quency Sound Reproduction,” J.
Audio Eng. Soc., 32:5, pp. 340-352
(May 1984).
[7] Flanagan, C., Wolf, R., and
Jones, W., “Modern Theater Loud-
430
speakers and Their Development,”
SMPTE (March 1937).
[8] Frayne, J. and Locanthi, B.,
“Theater Loudspeaker System Incor-
porating an Acoustic Lens Radiator,”
SMPTE, 63:3, pp. 82-85 (September
1954).
[9] Green, I. and Maxfield, J., “Pub-
lic Address Systems,” Bell System
Technical Journal, 2:2, p. 113 (April
1923). Reprinted in J. Audio Eng.
Soc., 25:4, pp. 184-195 (April 1977).
[10] Hilliard, J., “An Improved The-
ater-Type Loudspeaker System,” J.
Audio Eng. Soc., 17:5, pp. 512-514
(Oct. 1969). Reprinted in JAES, Nov
1978.
[11] Hilliard, J., "A Study of The-
ater Loudspeakers and the Resultant
Development of the Shearer Two-Way
Horn System," SMPTE, pp. 45-59
(July 1936).
[12] Hilliard, J., “Historical Review
of Horns Used for Audience-Type
Sound Reproduction,” J. Acoust. Soc.
Am., 59:1, pp. 1-8 (January 1976).
[13] Keele, D., “An Efficiency Con-
stant Comparison between Low-Fre-
quency Horns and Direct Radiators,”
presented at the 54th AES Conven-
tion, Los Angeles, 4-7 May 1976;
preprint 1127.
[14] Kock, W. and Harvey, F.,
“Refracting Sound Waves,” J. Acoust.
Soc. Am., 21:5, pp. 471-481 (Septem-
ber 1949).
[15] Lansing, J., “New Permanent
Magnet Public Address Loudspeaker,”
SMPTE, 46:3, pp. 212 (March 1946).
[16] Lansing, J. and Hilliard, J., “An
Improved Loudspeaker System for
Theaters,” SMPTE, 45:5, pp. 339-349
(November 1945).
[17] Locanthi, B., “Application of
Electric Circuit Analogies to Loud-
speaker Design Problems,” IRE Trans-
actions on Audio, volume PGA-4,
March 1952. Reprinted in J. Audio
Eng. Soc., 19:9, pp. 778-785 (1971).
[18] Martin, D., “Speaker Technol-
ogy for Sound Reinforcement,” Studio
Sound, March 1976.
[19] Novak, J. “Performance of
Enclosures for Low-Resonance, High
Compliance Loudspeakers,” J. Audio
Eng. Soc., 7:1, pp. 29-37 (January
1959).
[20] Olson, H., “Horn Loudspeak-
ers, Part I. Impedance and Directional
Characteristics,” RCA Review, vol. I,
no. 4 (1937).
[21] Olson, H., “Horn Loudspeak-
ers, Part II. Efficiency and Distortion,”
RCA Review, vol. II, no. 2 (1937).
[22] Rice, C. and Kellogg, E.,
“Notes on the Development of a New
Type of Hornless Loudspeaker,”
Transactions, AIEE, volume 44, pp.
982-991 (September 1925). Reprinted
J. Audio Eng. Soc., 30:7/8, pp. 512-
521 (July/August 1982).
[23] Thiele, N. and Small, R., Direct
Radiator Sealed Box, Vented Box, and
Other Papers Collected in AES Loud-
speaker Anthologies, Volumes 1, 2,
and 3, Audio Engineering Society,
New York, 1978, 1984, 1996.
[24] Thrasher, F., ed., Okay for
Sound . . . How the Screen Found Its
Voice, Duell, Sloan, and Pearce, New
York, 1946.
[25] Villchur, E., “Problems of Bass
Reproduction in Loudspeakers,” J.
Audio Eng. Soc., 5:3, pp. 122-126
(July 1957).
[26] Villchur, E., “Revolutionary
Loudspeaker and Enclosure,” Audio,
vol. 38, no. 10 (October 1954).
[27] Wente, E. and Thuras, A.,
“Auditory Perspective—Loudspeakers
and Microphones,” Electrical Engi-
neering, vol. 53, pp. 17-24 (January
1934). Also, BSTJ, XIII:2, p. 259
(April 1934), and J. Audio Eng. Soc.,
volume 26, number 3 (March 1978).
Direct Radiators
[28] Badmaieff, A., “Sound Repro-
ducing Device,” Altec “Duocone,” U.
S. Patent 2,834,424, issued 13 May
1958; filed 26 January 1956.
[29] Bank, G., and Harris, N., “The
Distributed Mode Loudspeaker—The-
ory and Practice,” AES UK Confer-
ence, London (16-17 March 1998).
[30] G. Beers and H. Belar, “Fre-
quency Modulation Distortion in
Loudspeakers,” Proceedings, IRE,
volume 31, number 4 (April 1943).
Reprinted in J. Audio Eng. Soc., 29:5,
pp. 320-326 (May 1981).
[31] Benson, J. E. “Theory and
Design of Loudspeaker Enclosures,”
Amalgamated Wireless Australia
Technical Review (1968, 1971, 1972).
[32] Button, D., “Heat Dissipation
and Power Compression in Loud-
speakers,” J. Audio Eng. Soc., 40:1/2,
pp. 32-41 (January/February 1992).
[33] Gander, M., “Moving-Coil
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April
 HISTORICAL PERSPECTIVES AND TECHNOLOGY OVERVIEW
OF LOUDSPEAKERS FOR SOUND REINFORCEMENT
Loudspeaker Topology and an Indica-
tor of Linear Excursion Capability,” J.
Audio Eng. Soc., 29:1, pp. 10-26 (Jan-
uary/February 1981).
[34] Gander, M., “Dynamic Linear-
ity and Power Compression in Mov-
ing-Coil Loudspeakers,” J. Audio Eng.
Soc., 34:9, pp. 627-646 (September
1986).
[35] Harris, N., and Hawksford, O.,
“The Distributed Mode Loudspeaker
as a Broad-Band Acoustic Radiator,”
presented at the AES 103rd Conven-
tion, New York, September 1997,
preprint 4526.
[36] Henricksen, C., “Heat Transfer
Mechanisms in Loudspeakers: Analy-
sis, Measurement, and Design,” J.
Audio Eng. Soc., 35:10, pp. 778-791
(October 1987).
[37] Hunt, F., Electroacoustics, J.
Wiley & Son, New York (1954).
Reprinted by the American Institute of
Physics for the Acoustical Society of
America, 1982, p. 59.
[38] Keele, D. B. (Don), “Maximum
Efficiency of Direct Radiator Loud-
speakers, 91st AES Convention, New
York, October 1991; preprint 3193.
[39] Klapman, “Interaction
Impedance of a System of Circular
Pistons,” J. Acoustical Society of
America, vol. 11, pp. 289-295 (Jan-
uary 1940).
[40] Klipsch, P., “Modulation Dis-
tortion in Loudspeakers: Parts 1, 2,
and 3,” J. Audio Eng. Soc., 17:2, pp.
194-206 (April 1969); 18:1, pp. 29-33
(February 1970); 20:10, pp. 827-828
(December 1972).
[41] Leach, M., “Electroacoustic-
Analogous Circuit Models for Filled
Enclosures,” J. Audio Eng. Soc.,
37:7/8, pp. 586-592 (July 1989).
[42] Olson, H., Acoustical Engi-
neering, D. Van Nostrand, New York,
1957. Reprinted by Professional
Audio Journals, Philadelphia, PA,
1991.
[43] Olson, H., “Gradient Loud-
speakers,” J. Audio Eng. Soc., 21:2,
pp. 86-93 (March 1973).
[44] Roozen, N. B., et al., “Vortex
Sound in Bass-Reflex Ports of Loud-
speakers, Parts 1 and 2,” J. Acoust.
Soc. Am., vol. 104, no. 4, (October
1998).
[45] Salvatti A., Button, D., and
Devantier, A., “Maximizing Perfor-
mance of Loudspeaker Ports,” pre-
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April
sented at the 105th AES Convention,
San Fransisco, 1998; preprint 4855.
[46] Siemens, E., U.S. Patent
149,797 (1874).
[47] Small, R., “Closed-box Loud-
speaker Systems, Parts 1 and 2.” J.
Audio Eng. Soc., 20:10, pp. 798-808
(December 1971) and 21:1, pp. 11-18
(January/February 1972).
[48] Strahm, C., “Complete Analy-
sis of Single and Multiple Loud-
speaker Enclosures,” presented at the
81st AES Convention, Los Angeles,
12-16 November 1986; preprint 2419.
[49] Thiele, N., “Loudspeakers in
Vented Boxes, Parts 1 and 2,” J.
Audio Eng. Soc., 19:5 and 6, pp. 382-
392, 471-483 (May and June 1971).
[50] Vanderkooy, J., “Loudspeaker
Ports,” presented at the 103rd AES
Convention, New York, September
1997; preprint 4523.
[51] Vanderkooy, J., “Nonlinearities
in Loudspeaker Ports,” presented at
the 104th AES Convention, Amster-
dam, The Netherlands, May 1998;
preprint 4748.
[52] Wolff, I., and Malter, L.,
“Sound Radiation from a System of
Circular Diaphragms,” Physical
Review, vol. 33, pp. 1061-1065 (June
1929).
[53] Zacharia, K. and Mallela, S.,
“Efficiency of Multiple-Driver
Speaker Systems,” presented at the
IREE (Australia) Convention, 1975.
Horns and Compression Drivers
[54] Frayne, J. and Locanthi, B.,
“Theater Loudspeaker System Incor-
porating an Acoustic Lens Radiator,”
SMPTE, 63:3, pp. 82-85 (September
1954).
[55] Geddes, E., “Acoustic Wave-
guide Theory,” J. Audio Eng. Soc., 37:
7/8, pp. 554-569 (July/August 1989).
[56] Geddes, E., “Acoustic Wave-
guide Theory Revisited,” J. Audio
Eng. Soc., 41:6, pp. 452-461 (June
1993).
[57] Goldstein, S. and McLachlan,
N., “Sound Waves of Finite Ampli-
tude in an Exponential Horn,” J.
Acoust. Soc. Am., vol. 6, pp. 275-278
(April 1935).
[58] Hanna, C., and Slepian J., “The
Function and Design of Horns for
Loudspeakers,” Transactions, AIEE,
vol. 43, pp. 393-404 (February 1924);
also, abridged text in J. AIEE, vol. 43,
pp. 250-256 (March 1924); discussion
pp. 1191-1197. Reprinted in J. Audio
Eng. Soc., 25:9, pp. 573-585 (Septem-
ber 1977) and volume 26, number 3
(March 1978).
[59] Henricksen, C. and Ureda, M.,
“The Manta-Ray Horns,” J. Audio
Eng. Soc., 26:9, pp. 629-634 (Septem-
ber 1978).
[60] Keele, D., “Low-Frequency
Horn Design Using Thiele-Small
Parameters,” presented at the 57th
AES Convention, Los Angeles, 10-13
May 1977; preprint 1250.
[62] Keele, D., “What’s So Sacred
About Exponential Horns,” presented
at the 51st AES Convention, Los
Angeles, 13-16 May 1975; preprint
1038.
[63] Keele, D., “Optimum Horn
Mouth Size,” presented at the 46th
AES Convention, New York, Septem-
ber 1973; preprint 933.
[64] Kinoshita, S. and Locanthi, B.,
“The Influence of Parasitic Reso-
nances on Compression Driver Loud-
speaker Performance,” presented at
the 61st AES Convention, New York,
November 1978; preprint 1422.
[65] Klippel, W., “Modeling the
Nonlinearities in Horn Loudspeakers,”
J. Audio Eng. Soc., 44:6, pp. 470-480
(June 1996).
[66] Klipsch, P., “A Low-Frequency
Horn of Small Dimensions,” J.
Acoust. Soc. Am., vol. 13, pp. 137-144
(October 1941). Reprinted in J. Audio
Eng. Soc., 27:3, pp. 141-148 (March
1979).
[67] Kock, W. and Harvey, F.,
“Refracting Sound Waves,” J. Acoust.
Soc. Am., 21:5, pp. 471-481 (Septem-
ber 1949).
[68] Leach, M., “A Two-Port Anal-
ogous Circuit and SPICE Model for
Salmon’s Family of Acoustic Horns,”
J. Acoust. Soc. Am., 99:3, pp. 1459-
1464 (March 1996).
[69] Leach, M., “On the Specifica-
tion of Moving-Coil Drivers for Low-
Frequency Horn-Loaded Loudspeak-
ers,” J. Audio Eng. Soc., 27:12, pp.
950-959 (December 1979). Com-
ments: J. Audio Eng. Soc., 29:7/8, pp.
523-524 (July/August 1981).
[70] Murray, F. and Durbin, H.,
“Three-Dimensional Diaphragm Sus-
pensions for Compression Drivers,” J.
Audio Eng. Soc., 28:10, pp.720-725
(October 1980). ➥
431
 HISTORICAL PERSPECTIVES AND TECHNOLOGY OVERVIEW
OF LOUDSPEAKERS FOR SOUND REINFORCEMENT
[71] Olson, H., “A New High-Effi-
ciency Theatre Loudspeaker of the
Directional Baffle Type,” J. Acoust.
Soc. Am., pp. 485-498 (April 1931).
[72] Olson, H., “Recent Develop-
ments in Theatre Loudspeakers of the
Directional Baffle Type,” SMPTE
(May 1932).
[73] Plach, D., “Design Factors in
Horn-Type Loudspeakers,” J. Audio
Eng. Soc., 1:4, pp. 276-281 (October
1953).
[74] Plach, D. and Williams, P.,
“Reactance Annulling for Horn Loud-
speakers,” Radio-Electronic Engineer-
ing, pp. 15-17, 35 (February 1955).
[75] Rocard, M., “Sur la Propaga-
tion des Ondes Sonores d’Amplitude
Finie,” Comptes Rendus, p. 161, 16
January 1933.
[76] Salmon, V., “Hypex Horns,”
Electronics, vol. 14, pp. 34-35 (July
1941).
[77] Salmon, V., “A New Family of
Horns,” J. Acoust. Soc. Am., 17:3, pp.
212-218 (January 1946).
[78] Small, R., “Suitability of Low-
Frequency Drivers for Horn-Loaded
Loudspeaker Systems,” presented at
the 57th AES Convention, Los Ange-
les, 10-13 May 1977; preprint 1251.
[79] Smith, D., Keele, D., and Ear-
gle, J., “Improvements in Monitor
Loudspeaker Design,” J. Audio Eng.
Soc., 31:6, pp. 408-422 (June 1983).
[80] Thuras, A., Jenkins, R., and
O’Neill, H., “Extraneous Frequencies
Generated in Air Carrying Intense
Sound Waves,” J. Acoust. Soc. Am.,
vol. 6, pp. 173-180 (January 1935).
[81] Webster, A., “Acoustical
Impedance and the Theory of Horns
and of the Phonograph,” Proceedings
of the National Academy of Sciences,
vol. 5, pp. 275-282 (May 1919).
Reprinted in J. Audio Eng. Soc.,
25:1/2, pp. 24-28 (January/February
1977).
[82] Wente, E. and Thuras, A., “A
High Efficiency Receiver for Horn-
Type Loudspeakers of Large Power
Capacity,” Bell System Technical
Journal, VII:1, p. 40 (January 1928).
Reprinted in J. Audio Eng. Soc., 26:3,
pp. 139-144 (March 1978).
[83] Wente, E. and Thuras, A.,
“Auditory Perspective—Loudspeakers
and Microphones,” Electrical Engi-
neering, vol. 53, pp. 17-24 (January
432
1934). Also, Bell System Technical J.,
XIII:2, p. 259 (April 1934). Reprinted
in J. Audio Eng. Soc., 26:7/8, pp. 518-
525 (July/August 1978).
[84] Wolff and Malter, “Directional
Radiation of Sound,” J. Acoust. Soc.
Am., vol. 2, pp. 201-241 (October
1930).
Arrays
[85] Augspurger, G. and Brawley,
J., “An Improved Collinear Array,”
presented at the 74th AES Conven-
tion, New York, 8-12 October 1983;
preprint 2047.
[86] Augspurger, G., “Near-Field
and Far-Field Performance of Large
Woofer Arrays,” J. Audio Eng. Soc.,
38:4, pp. 231-236 (April 1990).
[87] Davis., D and Wickersham, R.,
“Experiments in the Enhancement of
the Artist’s Ability to Control His
Interface with the Acoustic Environ-
ment in Large Halls,” presented a the
51st AES Convention, Los Angeles,
13-16 May 1975; preprint number
1033.
[88] Franssen, N., “Direction and
Frequency Independent Column of
Electroacoustic Transducers,” Philips
“Bessel” Array, Netherlands Patent
8,001,119, 25 February 1980; U.S.
Patent 4,399,328, issued 16 August
1983.
[89] Gander, M. and Eargle, J.,
“Measurement and Estimation of
Large Loudspeaker Array Perfor-
mance,” J. Audio Eng. Soc., 38:4, pp.
204-220 (1990).
[90] Keele, D., “Effective Perfor-
mance of Bessel Arrays,” J. Audio
Eng. Soc., 38:10, pp. 723-748 (Octo-
ber 1990).
[91] Kinsler, L., et al., Fundamen-
tals of Acoustics, third edition, Wiley,
New York, 1980.
[92] Kitzen, J., “Multiple Loud-
speaker Arrays Using Bessel Coeffi-
cients,” Electronic Components &
Applications, 5:4 (September 1983)
[93] Kleis, D., “Modern Acoustical
Engineering,” Philips Technical
Review, 20:11, pp. 309-348 (1958/59).
[94] Klepper, D. and Steele, D.
“Constant Directional Characteristics
from a Line Source Array,” J. Audio
Eng. Soc., 11:3, pp. 198-202 (July
1963).
[95] Kuttruff, H., Room Acoustics,
Applied Science Publishers, London,
1979.
[96] Meyer, D., “Multiple-Beam
Electronically Steered Line-Source
Arrays for Sound Reinforcement
Applications,” J. Audio Eng. Soc.,
38:4, pp. 237-249 (April 1990).
[97] Meyer, D., “Digital Control of
Loudspeaker Array Directivity,” J.
Audio Eng. Soc., 32:1, pp. 747-754
(1984).
[98] Rathe, E., “Note on Two Com-
mon Problems of Sound Reproduc-
tion,” J. Sound and Vibration, vol. 10,
pp. 472-479 (1969).
[99] Ureda, M., “Line Arrays: The-
ory and Applications," presented at the
110th AES Convention, Amsterdam,
May 2001; preprint 5304.
[100] Ureda, M., “‘J’ and ‘Spiral’
Line Arrays,” presented at the 111th
AES Convention, New York, Decem-
ber 2001; preprint 5485.
SUPPLEMENTAL
BIBLIOGRAPHY:
[101] Borwick, J., ed., Loudspeaker
and Headphone Handbook, third ed.,
Focal Press, Oxford, UK, 2001.
[102] Colloms, M., High Perfor-
mance Loudspeakers, fifth ed., John
Wiley & Sons, New York, 1997.
[103] Dickason, V., The Loud-
speaker Cookbook, sixth ed., Audio
Amateur Press, Peterborough, NH,
2000.
[104] Eargle, J., Electroacoustical
Reference Data, Van Nostrand Rein-
hold, New York, 1994.
[105] Eargle, J., Loudspeaker Hand-
book, 2nd ed., Kluwer Academic Pub-
lishers, Boston, 2003.
[106] Eargle, J. and Foreman, C.,
JBL Audio Engineering for Sound
Reinforcement, Hal Leonard Publica-
tions, 2002.
[107] Langford-Smith, F., ed.,
Radiotron Designer’s Handbook,
Amalgamated Wireless Valve Co,
Sydney, and Radio Corporation of
America, Harrison, NJ, 1953 (avail-
able on CD ROM).
[108] Merhaut, J., Theory of Elec-
troacoustics, McGraw-Hill, New
York, 1981.
[109] Olson, H. Solutions of Engi-
neering Problems by Dynamical
Analogies, second ed., D. Van Nos-
trand, Princeton, NJ, 1958.
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April
 Why advertise in the
Journal of the
Audio Engineering Society?
It's the best way to reach the broadest spectrum of
decision makers in the audio industry.

The more than 12,000 AES members and subscribers

worldwide who read the Journal include
record producers and engineers, audio equipment designers and
manufacturers, scientists and researchers,
educators and students (tomorrow's decision makers).

Deliver your message where you get

the best return on your investment.

For information contact:

Advertising Department, Flavia Elzinga
Journal of the Audio Engineering Society
60 E 42nd Street, Room 2520
New York, NY 10165-2520
Tel: +1 212 661 8528, ext. 34
journal_advertising@aes.org
www.aes.org/journal/ratecard.html
 DSP in Loudspeakers
By AES Staff Writer
Loudspeaker systems are getting clev-
erer thanks to the incorporation of
advanced digital signal processing
algorithms that can compensate for
some of the deficiencies in the trans-
duction process. Various forms of non-
linear distortion may be reduced or it
may be possible to get better perfor-
mance out of smaller units by using
electronics to counteract physical inad-
equacies. Some of these processes can
make use of psychoacoustical phenom-
ena, such as a means of extending the
perceived bass response without actu-
ally reproducing the relevant low fre-
quencies, and it may also be possible to
modify the way in which the loud-
speaker interacts with the listening
room. Finally there are various ways by
which it may be possible to engineer an
all-digital signal chain, even using digi-
tal forms of representation right up to
the point where the binary data is con-
verted into an acoustical waveform.
This article is based on a selection of
recent papers that were given at the
23rd AES International Conference on
Signal Processing in Audio Recording
and Reproduction, 23–25 May 2003.
The full proceedings of this confer-
ence can be ordered online at
www.aes.org/publications/conf.cfm.
DISTORTION COMPENSATION
Klippel considers the issues of distor-
tion compensation in his paper “Active
Fig. 1. General model describing the basic signal flow in loudspeakers using linear (thin) and nonlinear (bold) subsystems
(Figs. 1 and 2 courtesy Klippel).
434
Compensation of Transducer Nonlin-
earities.” He points out that such cor-
rection involves using DSP to deal with
what he calls “regular nonlinearities” as
opposed to mechanical defects in con-
struction (such as voice coil rubbing
and cone offset). The regular non-lin-
earities are those that would be present
in any loudspeaker with the same
design, owing to physical limitations
such as cone excursion, voice coil
length, and so forth. Such techniques
are increasingly required to produce
small loudspeakers with a high output
level, which can result in them being
driven into the large signal region
where linearity problems become more
noticeable. In discussing this issue,
Klippel says: “Active loudspeaker con-
trol can only be justified in applications
requiring smaller, less expensive trans-
ducers giving more output with lower
distortion at higher sensitivity.”
He goes on to describe the generic
characteristics of compensation sys-
tems that act to control the output of the
loudspeaker. Such controllers can be
based upon models of the loudspeaker
in question, so that they model more
precisely the known behavior of the
device. These, he points out, are to be
preferred to generic controllers and are
likely to be more successful. It is also
important that any control scheme
should be sufficiently flexible to adapt
to changes in conditions such as heat-
ing of the voice coil and long term
changes in loudspeaker parameters, and
that any sensor employed to monitor
the reproduced acoustic signal or cone
movement should be robust and stable.
Fig. 1 shows a generic model of the
various stages in the loudspeaker at
which nonlinearities can occur (shown
bold). Notice that only the amplifier
crossover and room acoustics are con-
sidered to be linear elements in the
chain, all other elements having some
degree of nonlinearity.
Using a signal processing software
module to compensate for physical
aspects of the loudspeaker linearity is
said to provide considerable new free-
dom for the designer because it now
becomes possible to consider loud-
speaker parameters with a different
order of priority. For example, size,
weight, and directivity can be chosen
more freely, assuming that distortions
arising from nonoptimal combinations
of such can be compensated for elec-
tronically. The hardware costs of
implementing such an approach may be
relatively small, perhaps requiring only
the addition of a current sensor.
As part of a detailed exposition of
techniques, he makes the point that
with analog linearization techniques
there cannot be a delay between the
loudspeaker input and the sensor out-
put, therefore a feed-forward adaptive
approach, such as that shown in Fig. 2,
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April
Fig. 2. Adaptive indirect control of a loudspeakers using a generic feed-forward structure both in controller and detector.

has to be adopted. The filter parameters
are adapted based on the loudspeaker
input signal u(t) and the acoustic output
p(t), making the system self-tuning.
However, the sensor can be deactivated
at any time and the filter will continue
to operate based on current parameters.
Bright, in “Simplified Loudspeaker
Distortion Compensation by DSP,” dis-
cusses similar ideas. He describes a rel-
atively straightforward concept in
which a discrete-time model of the
loudspeaker in question is inverted to
create an appropriate linearization filter.
He says, “the resulting algorithm for
compensation of nonlinear distortion is
relatively simple, consisting of one or
more second-order IIR filter blocks
(depending on the order of the linear
dynamics) and several polynomial eval-
uations.”
As in Klippel’s paper, Bright consid-
ers a feed-forward system, which needs
to be tuned in situ so as to take into
account specific conditions within the
loudspeaker in question such as long-
term drift, aging, and thermal changes.
This can be achieved with feedback
from an electrical current sensor. Over-
all this compensation enables manufac-
turers to use a shorter voice coil that can
be located in the most concentrated part
of the loudspeaker’s magnetic field,
having correspondingly lower mass and
higher sensitivity. While this is nor-
mally not done because of the nonlin-
earity that results at high voice-coil dis-
placements, the signal processing
described makes it possible.
Experimental results demonstrated
increased loudspeaker sensitivity com-
pared with traditional designs, owing to
the use of shorter voice coils, together
with a reduction in distortion compared
with the uncompensated case as shown
in Fig. 3.
PSYCHOACOUSTIC LF
EXTENSION
Aarts, in “Applications of DSP for
Sound Reproduction Improvement,”
describes a novel means for giving the
impression that a loudspeaker is
reproducing more bass energy than it
actually is. This is based on the psy-
choacoustic principle of the missing
fundamental or “residue pitch,” in
which the hearing mechanism tends to
assume the presence of a fundamental
frequency when harmonics of that ➥

Fig. 3. Bright’s results from one experimental loudspeaker with a shortened voice coil showing the distortion of an 800-Hz tone.
The left plot shows the acoustic pressure at 28 cm from the loudspeaker, and the right plot shows the amplifier’s output voltage.
The darker line shows the result with distortion compensation active (courtesy Bright).

J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April 435

 DSP in Loudspeakers
the possibility that intermodulation dis-
tortion may cause audible artifacts, and
that the added harmonics may interfere
with the existing sound spectrum
resulting in some timbral modification.

ALL-DIGITAL SIGNAL CHAIN

Tatlas et al., in “Towards the All-Digi-
tal Audio/Acoustic Chain: Challenges
and Solutions,” describe various issues
associated with the construction of an
all-digital signal chain from source to
loudspeaker, looking at ways in which
Fig. 4. Proposed implementation scheme for loudspeaker equalization in a room digital signals can be used directly to
(Figs. 4, 5, and 6 and Table 1 courtesy Tatlas et al.). drive loudspeakers.
After considering various aspects of
DSP DAMP DAE wireless networking and data reduction,
the authors consider loudspeaker–room
equalization, which is also covered later
in this article. According to Tatlas et al.,
PCM Q PCM Demux
“The all-digital transducer should incor-
(n/s, o/s)

N
porate sufficient DSP power, not only to
N’ (N’<N)
fs
Rfs realize simple functions such as volume
A control, delay, and shelving filtering, but
also to implement transducer response
DSP
correction and adaptation to its local
DAMP DAE
acoustic environment via the use of
equalization methods. These can incor-
porate FIR filtering on PCM data by
Q
PCM
(n/s, o/s)
PCM PCM to PWM Buffer
using room response inverse filters
N
fs
N’ (N’<N)
Rfs
1
2(2N-1)Rf s
derived from responses individually
B measured for each of the channel
receivers.” A possible implementation
DSP DAMP DAE scheme is shown in Fig. 4. However, as
we will see in the next section, this may
not always be the most appropriate way
of dealing with the problem of the
loudspeaker–room interface.
S/D
PCM
(n/s, o/s) Buffer Tatlas et al. go on to discuss digital
N
loudspeakers, explaining that current
fs 1
Rfs research is centered on two different
approaches—digital transducer arrays
C
(DTA) and multiple voice coil digital
Fig. 5. Three different approaches to the digital transducer array. (A) PCM; (B) PWM; loudspeakers (MVCDL). They prefer to
(C) 1-bit sigma–delta modulation. concentrate on the former for their anal-
ysis owing to promising characteristics
frequency are present even if the fun- frequency audio is also present. of the technology. DTAs are presented
damental is not physically present. There are various advantages and and tested with a variety of different dig-
Small loudspeakers tend to suffer disadvantages to this idea. The advan- ital audio signal types including multibit
from a poor low-frequency response, tages Aarts cites are: little energy is PCM, PWM, and 1-bit sigma–delta
making them ideal candidates for some radiated below the loudspeaker’s cut- (otherwise known as Direct Stream
form of enhancement. In this psychoa- off frequency; less headroom is Digital in the Sony/Philips SACD
coustic bass-enhancement system syn- required compared with more tradi- paradigm).
thetic harmonics of the “missing” bass tional “bass boost” approaches for a A typical digital transducer array, as
(that which the loudspeaker cannot comparable bass effect; the system is shown in Fig. 5, consists of three stages:
reproduce satisfactorily) are derived computationally and power efficient; it DSP (digital signal processing), DAMP
from the audio signal and added to the can be implemented with a simple ana- (digital amplification), and DAE (digital
part of the spectrum where the loud- log circuit if required; and it can also be audio emission).
speaker does radiate well, thereby giv- tuned to any kind and size of loud- The authors show how a typical
ing the impression that the lower- speaker. However, drawbacks include PCM digital transducer array might be
436 J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April
 DSP in Loudspeakers
Table 1. Theoretical (A) and actual (B) characteristics of for the possibility
the arrays tested. N is the resolution of the digital signal of obtaining any refer-
and R is the oversampling ratio employed. ence response by this
Digital Transducer Transducer
method. In other words,
Speaker Number Frequency the designer is able to
PCM 2N’-1 Rfs
ensure that the loud-
speaker sounds similar
PWM 2(2N’-1) Rfs
wherever it is placed in
SDM R fs different listening
A rooms.
Measurement of radi-
Digital Transducer Transducer ation resistance requires
Speaker Number Frequency
that measurements are
PCM 63 44–176 kHz taken at two different
PWM 510 44–176 kHz distances from the loud-
Fig. 6 . Simulated 6-bit PCM-based SDM 32 44–176 kHz speaker, using a method
array showing transducer matrix; 1 similar to that shown in
corresponds to the MSB of the digital B
Fig. 8. (A commercial
signal.
implementation of this
constructed in a matrix of small drivers not necessarily optimal despite the fact principle has a motorized microphone,
as arranged in Fig. 6, with each driver that it might intuitively seem so. mounted below the woofer, which pops
having a certain “weight” and driven ABC concentrates on the bass fre- out to two different distances during the
by a specific bit of the digital signal. quencies (20 to 500 Hz), views the calibration phase.) Once the measure-
The authors simulated different types problem from the loudspeaker’s situa- ments are available, a smoothed 16th
and topologies of these arrays, in an tion, and works by examining the radi- order IIR filter is calculated that
attempt to evaluate their relative perfor- ation resistance of the loudspeaker attempts to approximate the desired
mances. The three arrangements used (which is related to the power output). correction. ➥
are described in Table 1. The radiation resis-
Overall it was found that the PWM tance is affected by
version required an impractically large the existence of room
number of drivers as well as showing modes so this method
higher distortion than the other array makes it possible to
types. Oversampled low-bit PCM compensate for the
arrangements showed promising effect of these on the
results with a manageable array size. perceived loud-
The SDM version performed well on- speaker timbre,
axis and its physical characteristics adjusting for the
made it a practical proposition, but its loudspeaker’s current Fig. 7. The basic principle of the ABC system (Figs. 7 and 8
directivity characteristics would position in the room. courtesy Pedersen).
require some attention. Generally, the (For example, the
smaller sized DTAs were found to be bass sound pressure
preferable and the distortion results level might be up to 9 dB higher when
measured off-axis were considerably the loudspeaker is placed in a corner
poorer than those on-axis. when compared to the free field.) By
controlling the acoustic power of the
LOUDSPEAKER–ROOM loudspeaker in this way it is claimed
INTERFACE EQUALIZATION that the equalization of timbre is per-
Pedersen, in “Adaptive Bass Control— ceived at any point in the listening
The ABC Room Adaptation System,” room, rather than being strongly
points out that traditional loudspeaker- affected by changes in listening posi-
room equalization systems concentrate tion (see Fig. 7).
on measuring the transfer response to a The aim of ABC is to measure the
certain listening position and creating loudspeaker radiation resistance in a
an inverse filter. This has various prob- reference position in a reference room
lems including gain differences of and then in the target room, using the Fig. 8. The ABC system embedded
maybe 20 to 30 dB, sensitivity to ABC filter to equalize the resulting into an active loudspeaker. The
changes in listener position (coloration, response so that the target timbre is microphone is mounted on a vertical
rod, which by rotation through 180
preechoes), and nonminimum phase close to the reference timbre. This degrees effectively moves the
components in the filter. Furthermore, a assumes that the reference response microphone to a position 4 cm further
constant amplitude transfer function is timbre is the desired one, but allows away from the diaphragm.
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April 437
 DSP in Loudspeakers
Wilson et al., in “The Loud- ceived sound of the loudspeaker in
speaker-Room Interface—Con- the room and the relative impor-
trolling Excitation of Room tance of early and late sound. A
Modes,” discuss a similar con- tentative proposal is offered that
cept. They reinforce the points psychoacoustic evidence would
about the problems of equalizing suggest it is dangerous to introduce
the magnitude response at the lis- filters that have a notch greater
tening position, and they also than 6 dB (which therefore reduce
claim that it is preferable to con- the decay time of a room mode by
sider the way in which the loud- more than half), but this needs
speaker interacts with room more testing to verify.
modes. They summarize some of Waterfall plots from experi-
the problems noted in relation to mental work suggest a smoothed
loudspeaker-room mode interac- modal response resulting from
tion as follows: this system, as shown in Fig. 10.
1. Some frequencies are empha- Measurements of mode decay
sized or deemphasized, therefore time also suggest that the effect is
some notes in a series are too loud successful over a range of mea-
or missing. suring positions.
2. Some notes overhang or ring Fig. 9. System for measuring and controlling room Karjalainen et al. continue this
on much longer than others, con- modes (Figs. 9 and 10 courtesy Wilson et al.). theme in “Modal Equalization by
tributing to their dominance. Temporal Shaping of Room
3. The pitch changes during the the room between 500 and 2000 Hz. Response.” They too work on the prin-
decay of a note. Such correction can cause distortions ciple that traditional magnitude equal-
4. The pitch of short notes changes, to the direct sound signal from the ization (such as third-octave graphic EQ
such that the pitch heard is different loudspeaker, and so the question arises or basic loudspeaker equalization) can-
from the original. as to what primarily governs the per- not serve to control modal decay time
5. Echoes occur where a single tone
burst or note is changed to two or more,
shorter, tone bursts.
The energy decay rate at modal fre-
quencies tends to be longer than at non-
modal frequencies, causing overhang
and other nasty effects. One of the aims,
therefore, of the approach they describe
appears to be to create a filter such that
the combined response of the filter and
mode has a shorter decay time.
In order to devise the filters needed
to undertake this equalization, a mea-
surement to identify room modes is
required, finding those with the longest A
decay time (see Fig. 9). The authors
describe a novel method of achieving
this that attempts to overcome the prob-
lem that the measuring microphone
may be near a point of minimum SPL
in the modal response at the measuring
position. Using a long Hanning win-
dow, the impulse response is trans-
formed into the frequency domain and
even the smallest peaks are registered
as potential candidates for modes.
Information about the amplitude and
RT60 (reverberation time) of the modes
is used in a weighted fashion to deter-
mine the most critical modes for equal- B
ization. Most emphasis is placed on
RT60. Filters are then calculated with a Fig. 10. Waterfall plot showing room modes excited by the left loudspeaker (A). Plot
target RT based on the average RT in showing controlled excitation of modes from the same loudspeaker (B).
438 J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April
 DSP in Loudspeakers

A
Fig. 12. Decay envelopes (Schroeder-integrated) for the
original impulse response (upper solid line), for the AMK
equalized (dotted line), for the ARMA equalized (dashed
line), and for the windowing mode equalized response
(lower solid line).

Fig. 13. Decay time as function of frequency for original

Fig. 11. Waterfall plots showing initial (A) and reduced decay rate response (dash-dot line), the target curve (dashed line),
after modal equalization (B) with AMK filter (Figs. 11, 12, and 13 and for the equalized response (solid line) using the
courtesy Karjalainen et al.). windowing technique.

or the detailed structure of modal
response. So they resort to more sophis-
ticated forms of digital filtering in the
time domain that attempt to alter the
detailed modal response of the loud-
speaker–room combination.
The principle of the approach is
based on frequency-dependent win-
dowing of the impulse response of the
system. Essentially the impulse
response at the listening position is
measured and the decay time deter-
mined at each frequency, then the
impulse response is filtered with an
exponentially decaying function at
those frequencies where it is deemed
necessary to reduce it. This is achieved
by means of a relatively high-order FIR
(finite impulse response) filter. They
described two different filtering meth-
ods, known as AMK and ARMA, in
previous papers, the second of which
they say is better suited to the fitting of
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April
closely spaced modes and does not
require the prior estimation of modal
decay rate.
Fig. 11 shows the effect of such an
AMK filter, based on synthetically cre-
ated room modes at frequencies of 50,
55, 100, 130, and 180 Hz. The corre-
sponding decay times were given as
1.4, 0.8, 1.0, 0.8, and 0.7 s, and the
modal equalizer design target was to
reduce those decay times to 0.30, 0.30,
0.26, 0.24, and 0.20 s (some LF rise is
allowed as this is normal within room
acoustics standards). The authors note
that both types of filter tend to reduce
the initial decay rate quite well but that
then the decay slips back to being
closer to the original (as noted at the
lower frequency end on these graphs),
particularly when the modes are closely
spaced (for instance at 50 and 55 Hz as
here).
Looking at the overall decay rate
below 200 Hz (see Fig. 12), it was
found that the new windowing method
was more effective then either of the
original two methods, especially if no
hand-tuning was used with the AMK
and ARMA methods.
The plot of decay time versus fre-
quency (see Fig. 13) shows that the
decay time has been reduced success-
fully using the new windowing filter at
most frequencies, close to the target
maximum (dashed line).
It is clear from the examples pre-
sented in the papers summarized here
that the use of DSP in loudspeakers
can be an effective tool to counteract
inadequacies in their physical charac-
teristics and in their interaction with
the room. It may also be that all-digi-
tal reproduction chains will one day
become a reality if the problems asso-
ciated with digital transducer arrays
can be solved.
439
 SURROUND LIVE!
SUMMARY
Technology Visions, Overland Park, KA, USA
n October 9,
O 2003 a unique
day-long event
took place at the
Manhattan Center Studio
complex’s Grand Ballroom.
Surround Live was the first-
ever comprehensive event
devoted exclusively to the
creation, production, and reproduction
of live performance audio in multi-
channel surround.
Surround Live brought together
nearly 250 working professionals from
a cross section of the audio industry in
a one-day interactive workshop, in
conjunction with the AES 115th
Convention in New York, to discuss
the issues and technological challenges
created by presenting music, drama,
and theater in full multichannel
surround audio formats to a live
audience.
In an interesting piece of historical
juxtaposition, the Manhattan Center
complex is adjacent to and connected
by passageways to what is now the New
Yorker Hotel, which was the site of the
very first New York AES Convention
more than a half century ago.
The Surround Live event was orga-
nized into several distinct sections,
with formal presentations and discus-
sions taking up the morning and a
portion of the time after lunch.
The event then segued into a series
of demonstrations followed by a 90-
minute segment with a live band on
stage, showcasing various mixing and
music-presentation ideas and concepts.
Legendary FOH mixing engineer
Buford Jones—whose 30-year career
has included tours as a live-sound
mixer or engineer for such luminaries
as Stevie Wonder, Eric Clapton, Pink
Floyd (including a Platinum Record
for “Delicate Sound of Thunder”),
David Bowie, Faith Hill, Jeff Beck,
George Harrison, James Taylor,
440
Frederick Ampel
Surround Live panelists: clockwise
from above, Fred Ampel, Kurt Graffy,
Kurt Eric Fischer, Bruce Olsen, and
Steve Schull.
SHeDAISY, ZZ Top, and many
others—handled the console and the
external signal processing. A lively
and intense interaction ensued between
the attendees, the band, and the mixing
team, discussing various ways to
present the music experience and vari-
ous ways to use electronic reverb and
spatial effects, and then testing those
concepts live with the band while
listening and evaluating the results
immediately.
The live music performance was
provided by the Surround Live
Performers featuring guitarist Jeff
Golub (GRP Records, www.jeffgolub.
com).
The formal presentation portion of
Surround Live was opened by Kurt
Graffy of Arup Acoustics, San
Francisco. Graffy challenged
the attendees to start thinking
about physioceptualsophicacous-
tics, a term he created to encom-
pass the many complex aspects
of defining and understanding
the multichannel audio environ-
ment and its physical and
psychoacoustic aspects. His presenta-
tion first explored some of the previous
literature on the topic, including contri-
butions from Durand Begault on 3-D
sound, Jens Blauert on spatial hearing,
the extensive literature published by
David Griesinger, Brian C. J. Moore’s
work on masking, David R. Perrott’s
papers on audio and visual localization
and modalities, J. Robert Stuart’s
defining work on the psychoacoustics
of multichannel audio, Nick
Zacharov’s papers covering multichan-
nel level alignment, and the AES 16th
Conference in 1999 in Rovaniemi,
Finland on spatial sound reproduction,
as well as other papers from the AES
8th, 12th, and 15th Conferences. He
then conducted a review and refresher
for attendees on the physiological
aspects such as head shape, torso,
pinnae; perceptual aspects including
loudness, pitch, localization, envelop-
ment; the philosophical aspects includ-
ing reality modes, visual-auditory
modalities; the real-world acoustical
aspects such as spectrum, level, signal-
to-noise, directivity, and reverberant
level.
His presentation then covered a
range of other topics related to multi-
channel audio including perceptual
environmental distortions, perception
and depth, and formats and presenta-
tions of sonic imaging.
After a very focused question-and-
answer discussion, Bruce Olson of
Olson Sound Design in Minneapolis,
presented a detailed look at theatrical
sound concepts and formats incorporat-
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April
ing materials supplied by John Leonard
of Aura Sound Design in London.
Issues raised by Olson included the
number of channels, loudspeaker loca-
tion and position, audience perception
of sound effects, and spatial recreation.
It is hoped that a more extensive discus-
sion of this topic will be part of
Surround Live 2004 at the AES 117th
Convention in San Francisco with live
demonstrations of many of the points
from Leonard.
Next the attendees heard from Kurt
Eric Fischer of Attic Sound Design. His
sound design and production credits
include Tell Me on Sunday (Kennedy
Center), Marty (Huntington Theater
Company), Dorian (Denver Center
Theater Company), Blue (Roundabout
Theater Company, Audelco Award),
Linda Eder (Carnegie Hall), Finian’s
Rainbow (Coconut Grove Playhouse,
The Cleveland Playhouse), Macbeth
(New York Shakespeare Festival), the
Broadway production and first national
tour of Rent, Nine (Eugene O’Neill
Theater), Jesus Christ Superstar (Ford
Center Theater), Fascinating Rhythm
and Judgment at Nuremberg
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April
(Longacre Theater), the world
premieres of Parade (Drama Desk
Nomination) and Whistle Down the
Wind, Sunset Boulevard, Jelly’s Last
Jam, and Nick and Nora. Fisher
expanded on many of the concepts
addressed by Olson, and added his prac-
tical understanding of production reali-
ties and the difficulties of achieving
repeatability for a long running show or
an event that moves from location to
location, such as an on-the-road
Broadway show or live music tour.
Interaction
among the
attendees and
the panelists was
very lively
throughout the
event.
Following Fisher was Steve Schull
of Acoustic Dimensions, Dallas,
Texas. Schull’s extensive background
includes production credits on Les
Miserables, Grand Hotel, Cats, Lena
Horne, Annie, Little Shop of Horrors,
The Real Thing, Sophisticated Ladies,
Dreamgirls, Fences, and The Rocky
Horror Show. He has designed or
worked on aspects of the preservation
and restoration of a number of historic
theaters including the New Victory
Theater, the New Amsterdam ➥
 SURROUND LIVE SUMMARY
Theater, the New York State Theater at
Lincoln Center, the Vivian Beaumont
Theater at Lincoln Center, and the
Opera House at the Brooklyn Academy
of Music. And he’s been involved in the
management of new construction of the
Lucille Little Performing Arts Center at
Transylvania University, the Royal
Shakespeare Theatre, the Gallagher-
Bluedorn Performing Arts Center for the
University of Northern Iowa, and the
new performing arts center for Florida
State University. Schull addressed the
still relatively new ideas of implement-
ing and using multichannel sound in
religious facilities and some of the
more unexpected benefits and prob-
lems that occur when surround sound
is deployed in these facilities. The
rapid growth in megachurchs (some
have seating for 3000) was a focal
point for the presentation. A very
active discussion followed Schull’s
talk, and it is planned that he will
return to the panel at Surround 2004
with recorded examples of many of the
ideas and concepts offered in New
York. Many attendees were surprised
to learn of the magnitude of the
megachurch market, in which many
3000-seat churches now have full-scale
rock-and-roll style flying sound
systems, large-screen video systems,
and demanding production require-
ments. A lively debate ensued on what
audiences actually expect or demand at
megachurch services.
After a lunch break Duncan
Crundwell spoke about image-place-
ment techniques and level-based local-
ization technologies and methods.
Crundwell, president of 1602 Group
LLC, Alexandria, Virginia, brought to
bear his specific expertise in advanced
music control systems from his 17
years at SSL in pro-audio technical
sales, worldwide customer support,
and IT management, and his work in
studio design with Munro Associates
in the UK. Several very interesting
demonstrations were included using
the T-Max hardware/software plat-
form, and these produced an active
debate among attendees and panelists.
The psychoacoustics of “fooling” the
ear into believing that sounds were in
motion provided a key discussion point
after this presentation.
Mark Hood of Echo Park Studios
followed with a detailed look into
442
merging multichannel sound and the
other aspects of a show, using his
experience as sound designer of
Cyberjam, which had just opened at
the Queens Theatre in London’s West
End. The production makes extensive
use of surround sound. Hood also used
examples from his work as sound
designer for blast! (London,
Broadway, and the U.S. Tour), and the
U.S. tour of Shockwave. He was
recently audio producer of the PBS
special programs “Robert Mirabal,
Music from A Painted Cave” and
“blast! An Explosive Musical
Celebration.” Both were mixed and
released in 5.1 audio format. In his
presentation he also linked the concepts
from live theater with recorded and
studio aspects of surround based on his
work with John Mellencamp, George
Benson, David Sanborn, John Scofield,
Zamfir, Glenn Branca, the Gizmos,
Suicide, the Cincinnati Pops, and
hundreds of others. And he added a
historical note to the proceedings when
he mentioned that he still owns the
matrix boxes to decode his quadra-
phonic vinyl projects.
Later the attendees “got ready for
some football,” from Randy Hoffner,
manager of technical planning for the
ABC Television Network. He began
his presentation by noting that while
others were dealing with potential
audiences of a few thousand people,
the Monday Night Football production
crew was supplying surround sound to
millions of television viewers every
Monday night. He highlighted the
advance of HDTV and 5.1 audio by
showing two photographs of the same
production location; one taken in 2000
with a huge analog production truck
and small auxiliary digital van, and the
2003 version with the gigantic
HDTV/surround audio production
vehicle all by itself at the stadium.
Now the events are produced in HDTV
and digital sound and down converted
for the analog feeds, the reverse of just
three years ago.
Hoffner provided D5/HDTV 5.1
surround tape of a MNF game that
showcased the dramatic shift from the
preproduced opening credits in 4x3
and analog stereo, to the dramatic
jump to 16x9 and full 5.1 surround as
the game went live on air. The effect
truly startled the audience and was
repeated several time by request. Many
questions about how things were done
live and how the production was
handled followed his presentation.
The formal presentations concluded
with a talk by Marvin Welkowitz from
the NYU School of the Performing
Arts, Music Technology Department.
He spoke on the “why” of surround.
Since much of the day had been spent
looking at the “what” and “how,”
Welkowitz provided a stimulating look
at what benefits surround can offer to
various styles of live events, how
sound levels can be better managed
when the audience is placed within the
sound field, and how the musical or
show experience can be improved for
everyone involved. He challenged both
the panel and the audience to think
about how we move ahead with these
concepts and how more audio reality
can be created. Both the attendees and
the panelists reacted strongly to his
ideas, which carried over into the ques-
tion-and-answer session that followed,
which continued until the start of the
live performance.
Surround Live was made possible
through an extensive technical support
infrastructure provided through the
generous support of Meyer Sound,
DiGiCo, Shure, and TC Electronics.
Very special thanks go to Tim
Chapman and the entire Meyer Sound
and Sound Associates crew, Nick
Woods of Shure, Bob Doyle and the
DiGiCo team, Ed Simeone and the TC
team, and the crew of professionals
from the Manhattan Center, without
whom the event would not have been
possible. The Surround Live concept
and content was created and the event
was produced by Frederick J. Ampel
of Technology Visions.
Additional special thanks go to the
AES and Roger Furness for allowing
the event of take place and supporting
the concept, and to Mark Herman,
Keith and Julie Clark of Live Sound
International, and ProSoundWeb for
their unwavering commitment and
promotional support. The next install-
ment of Surround Live, More
Surround! The Art of Multichannel
Live Audio, is tentatively scheduled for
the October 27, 2004, the day before
the opening of the AES 117th
Convention, October 28–31, 2004 in
San Francisco.
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April

Tom Holman (on the right) chats with Chicago Section perceptible improve-
members during November meeting.
Surround Sound
Tomlinson Holman spoke to Chicago
Section members on November 6,
about the history and future of sur-
round sound. He presented the history
of surround as an opera in many acts,
with each era of surround being an act.
Holman began by tracing the history
of surround sound from its inception
centuries ago in antiphonal choirs and
organs. He described the history of mul-
tichannel reproduction systems from the
1933 Bell labs demonstrations using 3-
channel telephony through current 10.2
channel experiments at TMH labs.
Holman then outlined the technical
capabilities of past film sound formats,
including Fantasound, Cinerama, Cine-
mascope, Todd AO, and several analog
Dolby formats on 35- and 70-mm film.
He provided informative and often
amusing descriptions of the rise and fall The Pacific Northwest Section’s Sep-
of these formats. He talked about the
rationale behind the selection of 5.1 dis- line Community College Music Build-
crete channels for digital film audio and ing in Shoreline, Washington, featured
how this has been extended to 6.1 and
7.1 channels through the use of matrix-
ing. He also touched on the history of
multichannel sound for home, from ear-
ly 4-channel analog matrixed sound on
LP to current surround sound digital
formats available on DVD.
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April
NEWS
OF THE
SECTIONS
We appreciate the assistance of the
section secretaries in providing the
information for the following reports.
According to Holman,
using additional bits to
increase the bandwidth,
sample rate or the num-
ber of bits per channel
beyond current standards
provides little or no
demonstrable improve-
ment in audio quality.
However, using these
same extra bits to
increase the number of
channels makes an easily
ment, even to untrained
listeners. Holman ex-
plained that it is nearly impossible to
ever have enough channels for full
transparency. Furthermore, increases in
the number of channels will continue to
provide improvements in realism.
Holman said that specific shortcom-
ings of the existing 5.1 channel system
include poor side images, poor sense
of envelopment—especially at low fre-
quencies; poor rear-center imaging and
no vertical information. A system with
more channels would address these
problems, but could still be down-
mixed to be compatible with current
systems having fewer channels. Data
to control the down-mix can be includ-
ed with the music data, resulting in a
scalable system. The meeting conclud-
ed with informal discussions.
Panel on Mastering
tember 30, meeting held at the Shore-
a panel discussion with six local mas-
tering engineers. Some 145 people—
possibly a record—gathered to hear a
discussion among panelists Ed Brooks
and Rick Fisher of RFI/CD Mastering
in Seattle; Mark Guenther of Seattle
Disc Mastering; Max Rose of Disc-
makers’ Seattle office; Al Swanson of
the Seattle Symphony and Location
Recording; and Steve Turnidge of
Ultraviolet Studios. Opening com-
ments were made by PNW committee-
man/moderator Bob Gudgel and
section chair Dave Tosti-Lane.
Gudgel directed the first question to
the panel: What is mastering? The
responses reflected the fact that it is no
longer composed of the steps needed
for simply preparing an LP pressing.
Some answers included:
Mastering is “shoehorning” the
material into the CD format if it is not
already in 16 bit/44.1 kHz;
Mastering is making a marketable
CD from material in various states of
polish;
Mastering is the last step to a fin-
ished product, during which all final
details must be made;
Mastering is determining and eradi-
cating the differences in classical
material—songs may not have to be
matched up, but other similar finish
details are made;
Mastering is like a polishing
process, or putting a package together
to make a homogeneous complete CD.
Panelists addressed the question of
whether mastering engineers typically
do remixing. The consensus was that
sometimes they do, or they may tell
the client to go back and remix if the
material is not yet acceptable. It was
noted that mastering is not like mix-
ing. Mastering engineers are some-
times involved in the sequencing of
tracks. The consensus was that little
automation is used for mastering.
After stories of ego clashes during
recording sessions, Fisher remarked
that mastering engineers have the won-
derful advantage of not knowing where
the “bodies are buried.” Their only
concern is the music. Panelists agreed
they have to be nonjudgmental about
the music—they just shine it up, ➥
443
 NEWS
OF THE
SECTIONS
Pacific Northwest Section
members during the
September meeting.
Photo by Gary Louie
no matter how good or bad.
Do mastering engineers like to have
the client on the premises during mas-
tering? Al Swanson noted that for
classical work, it was usually not a
good idea to have a client present at
the editing stage (what he called the
tree stage) but it is okay at the assem-
bly stage (or forest stage). Others rec-
ommended having the client there.
What makes an album finished
rather than just one with a group of
songs? A finished product must meet
the expectations of the consumers; for
example, when the volume level of
each song is even and one does not
have to dive for the volume knob.
What about the loudness wars of
CDs? Al Swanson noted that even
classical music has some issues with
wanting it all to be loud. Others com-
mented that most material that is
squeezed/compressed and loud arrives
for mastering like that. It is hard to
make mastering changes if there is no
headroom left with which to work.
Is there a preferred format to receive
material? The general agreement was
that 1⁄2-inch stereo analog sounds the
best, followed by 1⁄4-inch analog. The
exception was Al Swanson, who said
he never uses analog anymore. How-
ever, others will always use an analog
tape stage somewhere during master-
ing; it is akin to using a processor for
the sound, even when the recording
will end up digital.
What are the essential tools for mas-
tering? Most agreed that the most
important factor was a set of good
loudspeakers. Also discussed were the
room and signal path preferences.
Nearly everyone tried to steer clear of
the “magic bullet” gear syndrome, but
everyone had something to say about
the importance of the room and moni-
444
toring system. Fisher mentioned that
this is probably the only place where
professional audio meets the lunatic-
fringe hi-fi gear.
How does one master for end-user
playback environments, such as head-
phones, cars and living rooms? Swan-
son noted that there is simply no
standard for home environments. Inter-
estingly, no one had much insight into
this factor, although they noted that
CDs were good for client test copies,
as opposed to the old LP pressings and
cassette days.
After a break for refreshments, the
panelists played some mastering
examples on a Genelec system. Deliv-
ery formats used to send the master to
the pressing plant include: Audio CD-
R, although it is not the best choice
due to possible inaccuracy in data
retrieval; Exabyte DDP, a data tape
format that was once popular but is no
longer made; ftp (file transfer proto-
col), which sends audio files straight
to the plant via the Internet; and
PMCD – Pre-Master CD, or an audio
CD-R made to full Redbook Audio
standards. The term originally
described a proprietary format from
Sony and Sonic Solutions but is now
commonly understood as a general
media format.
Each panelist briefly described his
own mastering set-up. Ed Brooks said
he often converts material to analog
and uses outboard processing. Pan-
elists agreed that the material they
receive needs a little headroom to
work with. Normalized or clipped
material is not good and severely lim-
its the changes that can be made. This
led to a discussion of loudness on CDs
and broadcasts. One question asked
was, “What happens to my mix when
it gets played on the radio?” Mark
Guenther had a CD of examples from
Orban, manufacturers of the Optimod
signal processor used by most mid-
large market stations. He played
examples demonstrating different set-
tings of the device and how they
affected the music.
Regarding the question of multiband
compression, panelists said that they
did not generally use it because it is
hard to set up, but it depends on the
source. Finally, someone asked: How
does one select a mastering engineer?
The general advice was to be comfort-
able with the engineer.
Special thanks to the participating
mastering engineers who provided
insight into their craft: AES PNW
Committee member Gudgel, who
organized the meeting; Opus 4 Studios
for the Genelec monitor system; Unee-
da Audio for the sound reinforcement
system; and First Choice Marketing
and Mackie Designs for their dona-
tions of door prizes.
Gary Louie and Rick Chinn
A la Mode…
On December 9, members and guests
of the British Section gathered to hear
Rhonda Wilson and Michael Capp of
Meridian Audio talk about “The
Loudspeaker-Room Interface—Con-
trolling Excitation of Room Modes.”
Wilson began by stating the aim of
the paper: to identify the strongest
room modes excited by each loud-
speaker and to prefilter the signal fed
to each loudspeaker in such a way that
the adverse effects of dominant room
modes are significantly reduced. Such
a system must be spatially robust and
simple to set up. Improvements at one
listening position should create
improvements, not degradations, at
other listening positions.
Typically, customer room specifica-
tions constitute small room acoustics.
The reverberant field under these cir-
cumstances is very important. Wilson
discussed various methods of correc-
tion, as well as methods to avoid, such
as full bandwidth magnitude response
equalization and full bandwidth
impulse response equalization. These
methods are very sensitive to position
effects. Audio delay may be neces-
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April
 NEWS
OF THE
SECTIONS
sary, which is not appropriate for
DVD video playback.
Modal ranges cover 20 Hz to 200
Hz. Room effects include emphasis of
some frequencies, ringing at resonant
frequencies, pitch changes during
decay, and beat and echo effects. The
room resonant modes of rigid walled
rooms of certain dimensions can be
calculated. Usually, both fundamental
and harmonic modes are set up. Non-
rigid walls cause the modes to decay
exponentially. Wilson showed graphs
of the decay of room resonance, beat
frequency effects, and echo effects.
Although the magnitude of a room
mode varies with listening position,
the decay time of a room mode is the
same at different listening positions.
There is no requirement to use a spe-
cial microphone. Hence there is a
focus on decay time.
The filter to control decay time uses
original Q, target Q, and center fre-
quency. The resulting notch filter has
a gain and bandwidth that can be cal-
culated based on target decay time and
room mode decay time. Plots of level
against delay time after correction
showed some sensitivity to pole fre-
quency errors. To avoid errors, 32-bit
coefficients and double precision
implementations are also necessary.
Wilson said that it is not desirable to
create an anechoic environment. Pre-
vious work in the measurement of the
reverberation time of living rooms has
yielded times of 0.4 to 1 second, quite
long compared to studios.
Some work was done to make sure
that the direct response was not overly
corrupted, and plots showed that the
direct response changes phase when
the exciting signal is switched off.
Music is much more complex than
tones. All prefiltering is done below
250 Hz. The trade-off is to prefilter
enough to reduce the reverberation
without affecting the direct response
too much. Notches up to 6 dB deep are
okay. Larger notches are more
problematic.
Capp continued, saying that the stan-
dard process is used to characterize the
decay time in the range 500 Hz to 2
kHz and then to set a target decay time
for modes below 250 Hz, which
increases as frequency decreases. A
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April
third octave bandpass filter is used for
measuring the decay time, followed by
Schroeder integration. According to
Capp, a more accurate process is
required to determine decay of individ-
ual room modes. Waterfall FFT plots
do give a better resolution. The
Schroeder integration is prone to noise,
so special attention is given to starting
from the peak impulse and ending just
above the noise floor. As an example,
Capp showed several waterfall plots.
Least squares regression is used to give
the optimum decay time.
To identify the dominant room
modes, all the peaks in the magnitude
response must be measured. After the
decay time at each room mode is
found, the dominant modes are identi-
fied using the decay time. A more
sophisticated method is to calculate
many decay profiles from a waterfall
plot. All the decay profiles are then
summed and the largest peaks represent
the dominant modes. Example plots
clearly showed the dominant modes.
Capp described the system imple-
mented to make these measurements,
which includes a microphone or SPL
meter in the center of the room, a PC
and a surround decoder. An MLS test
signal is used for each loudspeaker,
from which the impulse response for
each loudspeaker is calculated. Then
the designer performs an FFT and
decay time analysis, followed by de-
sign of the filters for each loudspeaker.
Meridian has developed PC software
driven by a user-friendly wizard to per-
form the measurements and correction.
Capp discussed a sample room and
showed magnitude responses for dif-
ferent listening positions before correc-
tion. Postcorrection magnitude and
decay responses showed clear
improvement. The filters generated for
one listening position improved the
responses for many listening positions.
In conclusion, this new spatially
robust filter design technique uses only
one microphone position but controls
the decay time at multiple positions.
The filter design is automated and does
not require specialist knowledge.
According to Capp and Wilson, so far
initial feedback from dealers and cus-
tomers has been positive.
More information on this topic is
available in the paper, “The Loudspeak-
er-Room Interface—Controlling Excita-
tion of Room Modes,” by Rhonda J.
Wilson, Michael D. Capp, and J. Robert
Stuart, Meridian Audio, AES 23rd Int’l.
Conference, Copenhagen, 2003.
Steven Harris
Triple Treat in New York
Some 70 members and guests of the
New York Section met on December
8, at Innovative Audio’s high-end
showrooms on Manhattan’s East Side
for a holiday party and annual triple
treat meeting.
The evening’s format featured three
presenters, Thom Cadley, Malcolm
Addey, and Jim Anderson, who held
court to share their engineering skills.
Cadley played DVD-Audio 5.1 sur-
round mixes from his work with artists
Beyonce, Billy Joel, Stevie Ray Vaugh-
an, and others. Addey brought in a
beautifully spacious two-track remote
of a small string orchestra he recorded
in a West Side church. Anderson com-
pared four release versions of Patricia
Barber’s Companion album on CD,
XRCD, SACD, and vinyl.
After a brief welcome and introduc-
tion by emcee Allan Tucker, the partic-
ipants split up into three isolated sound
rooms to spend the first of three half-
hour sessions listening to the presenter
of their choice. After thirty minutes,
each participant switched to another
room to experience a different show.
In this way, everyone got to enjoy all
three programs.
The central meeting area provided a
perfect spot for relaxing, schmoozing,
and snacking. Conversations casually
overheard touched on topics such as
the placement of Blumlein micro-
phones, the difficult state of the
recording business, and postulations
on why each attendee could not go
home with a complimentary plasma
display. Maybe next year.
Allan Tucker
Webster Reports Growth
Students from the Webster Universi-
ty Student Section are pleased to
report that a number of new students
have applied for membership and ➥
445
 NEWS
OF THE
SECTIONS
that the section appears to be steadily
growing. The section welcomed some
of these new members at a meeting
held on January 13 at the university.
Section chair John Jory talked to the
32 students present and described both
the section’s achievements and hopes
for the upcoming semester. The stu-
dents worked on a radio spot to
announce the surround sound demon-
stration on February 17. Eric Black-
mer, chairman of Earthworks, agreed
to give a lecture to the group sometime
in the future. In addition, the section
plans to hold elections for new com-
mittee members. Officers will remain
in their positions for the next semester.
Andy Weidmann
Planar Transducers
The Los Angeles Section’s November
25 meeting was hosted by HPV Tech-
nologies in Costa Mesa. The company
develops, manufactures, and markets
concert loudspeaker systems using
planar transducers.
Before entering the meeting, each
person was required to sign a nondis-
closure agreement by company staff.
Dragoslav Colich began his talk
with the recent evolution of planar
transducers. He spoke of his own jour-
ney in audio transducer research and
development, including a description
of the magnet arrangements used in
such drivers.
HPV integrates these relatively
small, flat drivers into large arrays,
creating highly efficient yet remark-
ably lightweight systems. Colich ex-
plained the benefits and restrictions of
arraying the drivers. Two end users,
including the well-known event audio
specialist Gary Hardesty, described
their experience with using the arrayed
systems in situations ranging from the
Arco Arena in Sacramento to a Broad-
way on Times Square show, to an out-
door concert in Washington, D.C.
After the presentations came the
demonstration of what these loud-
speaker systems can do, with seg-
ments of jazz, pop, and classical
music. Some was quite loud and chest-
thumping—especially the subwoofer-
fed cannon shots in Tchaikovsky’s
1812 Overture Solenelle—yet devoid
446
of audible distortion.
Because the presentations and
demonstration ended late, the group
didn’t get a chance to hear the concert-
sized system that had been set up out-
doors.
Bob Lee
Disney Concert Hall
Thanks to the efforts of section trea-
surer Kahne Krause, the section
enjoyed a holiday concert on Decem-
ber 13 by the Canadian Brass at the
new Walt Disney Concert Hall. In a
pre-concert dinner at McCormick and
Schmick’s, Yasushisa Toyota, a prin-
cipal designer for Nagata Acoustics,
spoke to the group about the acousti-
cal design and lineage of the hall.
Toyota began by explaining the char-
acteristics of two leading types of con-
cert hall architecture: the shoebox and
the vineyard. He showed photos and
diagrams of halls representative of the
two philosophies. The shoebox, com-
pact and rectangular shape, offers excel-
lent acoustics to the listening
audience but is limited in usable size.
To accommodate larger numbers of
patrons, designers have increasingly
turned to the vineyard approach, so
named because the curving rows and
fractured seating sections as viewed
from the stage resemble the hills of a
vineyard.
According to Toyota, a hurdle to be
overcome in the vineyard design is the
issue of primary reflecting surfaces. In a
shoebox, the walls define the room
boundaries and are fairly close to the
listening audience, providing strong pri-
mary reflections to reinforce the direct
sound from the stage. Vineyard design-
ers incorporate numerous small, often
curving walls that not only separate the
seating sections from one another but
also serve as primary reflectors. The
Walt Disney Concert Hall is but the lat-
est example of this vineyard design.
For most of the AES contingent, this
concert was the first visit beneath and
within the wavy stainless steel of archi-
tect Frank Gehry’s conception, which
was certainly a treat. The Canadian
Brass, a quintet of tuba, trombone,
French horn, and two trumpets enter-
tained with a winning combination of
virtuoso playing and humor. The pro-
gram included sing-alongs of Christmas
songs, a condensed and minimalist per-
formance of Bizet’s Carmen, and a
medley of Hanukkah tunes that featured
Josef Burgstaller’s remarkable imitation
of a klezmer clarinetist, performed on a
piccolo trumpet.
The group thanked Kahne Krause
for putting the program together. She
did it almost single-handedly (with
some help from her husband Ira), han-
dling the tickets, reservations, dinner,
and guest speaker.
Penn State Wrap-up
The Penn State Section announces
membership growth due to the group’s
successful attempts to reach out to the
campus and community. The section
is now on a campaign to gain more
national members, as well as add more
variety to meeting topics.
On September 29, 30 people gathered
to hear Robin Miller of Filmmaker, Inc.
Miller gave an abbreviated version of
his talk for the 115th AES meeting in
New York. The presentation outlined
his patent-pending method of recording
3-D sound (PerAmbio 3D/2D) to more
completely capture the sound field
experienced by a listener in a live
acoustic space. These recordings can be
reproduced in 2-D on current 5.1/6.1
surround sound systems. The complete
3-D sound field can be reproduced with
the help of a decoder.
On October 7, nine members joined
life fellow and former AES governor
(1977-1979) Geoffrey Wilson in a dis-
cussion entitled “British Audio—75
Years Ago.” In his talk, Wilson remi-
nisced about his boyhood memories of
his father, Percy, who was technical
advisor to The Gramophone Magazine
from the late 1920s until the start of
World War II in 1939. Most of his
memories had to do with the large horns
that Geoffrey’s father kept around the
house, as well as his experiments with
early phonographs—many of which
were kept in their living room. The
meeting served as a wonderful excur-
sion into the history of audio.
Wilson showed many slides, most of
which were taken from a 1974 paper
“Horn Theory and the Phonograph,”
which historically detailed the devel-
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April
 NEWS
OF THE
SECTIONS
opment of the acoustic theory of
the phonograph.
Acoustics
On November 17, 23 members and
guests of the Penn State Section gath-
ered for David Swanson, Penn State
Acoustics faculty member .
Swanson gave an overview of some
topics to be covered in the upcoming
acoustics class, Advanced Digital
Audio Signal Processing. He briefly
described many digital audio formats
and algorithms, complete with exam-
ples, and touched on topics such as
digital filtering, audio effects, MPEG
2 Level 3 (MP3) and noise cancella-
tion. The entire presentation can be
found on the Penn State Web site at:
www.clubs.psu.edu/up/aes/.
On December 8, 22 members gath-
ered for the final meeting of the
semester to hold elections. The new
officers are as follows: Dan
Valente, chair; Alexandra Loubeau,
vice-chair; Mark Gramann, treasurer;
Remy Gutierrez, secretary; members-
at-large: Miguel Horta, Erin (Jinx)
Horan, Liz Cosharek.
Valente expressed the need for peo-
ple to be more active in the Society
and outlined a plan devised by the
officers to increase membership. At
the core of this plan is the need to gen-
erate more publicity for meetings and
provide variety in the subjects cov-
ered. Members discussed ideas for
upcoming meetings and former chair
Kevin Bastyr suggested that smaller
meetings should be held where mem-
bers give 10- to 15-minute talks or
lead discussions. This will help keep
members active. Brian Tuttle, another
former chair, suggested that members
should brainstorm a new project (simi-
lar to the PVC loudspeaker designs)
for the upcoming year.
Thanks to the new publicity cam-
paign, the first meeting of the spring
semester on January 19 was enor-
mously successful. The section kicked
off with Mike Sokol’s 2004 surround-
sound tour, which is being presented
by Fits & Starts Productions. Over 60
people showed up for this semi-
nar/workshop. Sokol described tech-
niques in surround sound engineering.
During the session, he gave examples
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April
of the artistry of choosing each loud-
speaker in a surround sound set-up. He
continued with a discussion of the dos
and don’ts of recording surround.
The audience was very interested in
the wide variety of choices an engineer
has in deciding how to mix music in a
surround environment. In the second
half of the three-hour seminar, Sokol
showcased some software and equip-
ment from his sponsors: Tannoy, SRS,
DTS, MaxVision, and Steinberg. In all,
it was a highly enjoyable meeting.
Dan Valente
Czech Seminar
Forty-five members, guests, and stu-
dents of the Czech Section attended a
day-long seminar held in the Congress
Hall of the Promopro Company of
Prague on December 8. During the
morning session, members from
Prague and Brno presented several
papers and held discussions and
demonstrations. The papers included:
“Present Trends in the Development of
Digital Signal Processing,” by Profes-
sor Zdenek Smekal; “Digital Signal
Processor for AM Transmitters,” by
Ing. Pavel Stranak; “Systems for Mul-
tichannel Sound Recording and Repro-
duction,” by Ing. Richard Jejkal; and
“Transducers Using Direct Digital-
Analog Conversion,” by Libor Husnik.
After the technical symposium and
lunch break, the section held a busi-
ness meeting. Chair Pavel Baladran
reviewed the events of the past year
and asked that he not be nominated to
continue as chair for the 2004 season.
He plans to retire and move. Secretary
Jiri Ocenasek discussed the financial
report and made several recommenda-
tions regarding member databases and
e-mail correspondence. There was also
some time spent on planning transport
and accommodation for those Czech
Section members who will travel to the
AES Convention in Berlin.
The election results for 2004 were:
Libor Husnik from the Technical Uni-
versity of Prague will be chair. He will
also maintain his role as faculty advi-
sor for the student section. Ing.
Zdenek Kesner will remain as vice
chair, and Ocenasek as treasurer.
Jiri Ocenasek
Upcoming Meetings
2004 May 8-11: 116th AES Con-
vention, Messe Berlin, Berlin,
Germany. Contact: e-mail:
116th_chair@aes.org. See
page 464 for details.
•
2004 May 17-21: International
Conference on Acoustics,
Speech, and Signal Process-
ing (ICASSP 2004), Montreal,
Canada. On the Internet: www.
icassp2004.com.
•
2004 June 17-19: 25th Interna-
tional Conference, London,
UK, “Metadata for Audio.”
Contact John Grant, chair, e-
mail: 25th_chair@aes.org.
•
2004 July 5-8: 11th International
Congress on Sound and
Vibration (ICSV11), St. Peters-
burg, Russia. Contact Con-
gress Secretariat at P. O.
Box 08A9, 1st Krasnoar-
meyskaya Str., 1, 190005, St.
Petersburg, Russia, or e-mail:
icsv11@rol.ru.
•
2004 July 12-14: Noise-Con
2004, Baltimore, MD, USA,
INCE/USA, e-mail: ibo@
inceusa.org.
•
2004 August 22-25: Internoise
2004: Prague, Czech Repub-
lic. On the Web: www.inter-
noise2004.cz.
•
2004 October 1-3: 26th Interna-
tional Conference, “High
Quality Analog Audio Pro-
cessing,” Baarn, “the Nether-
lands. See page 464.
•
2004 October 6-8: The NAB Ra-
dio Show, San Diego, CA,
USA. For information call 800-
342-2460 or 202-429-5419.
•
2004 October 28-31: 117th AES
Convention, San Francisco,
CA, USA. See page 464 for
details.
•
2004 November 15-19: 148th
Meeting of the Acoustical
Society of America, San
Diego, CA, USA, e-mail:
asa@aip.org.
447

Solid State Logic won a Technical Grammy at the 46th Awards ceremony on
February 8. Shown (left to right): Rich Plushner, president of SSL; Phil Ramone,
producer/engineer; and Phil Wagner, SSL’s senior vice president.
ABOUT PEOPLE…
AES sustaining member Klipsch
Audio Technologies has announced
that Jim Breen has joined the com-
pany as business unit manager of
Commercial Contracting and Corpo-
rate Development. In his new posi-
tion, Breen is responsible for the
overall direction and distribution of
Klipsch brand audio products to
hotels, restaurants, houses of wor-
ship, retail stores, theme parks, board
and conference rooms, and other
commercial establishments.
Over the past year Klipsch loud-
speakers have been installed in Gap,
Banana Republic, Old Navy, Ron
Jon’s, and Bloomingdale’s stores, as
well as at hundreds of BP gas sta-
tions across the country. Breen will
work to maintain a steady stream of
high profile installations and increase
Klipsch’s share of the commercial
contracting market.
Before joining Klipsch, Breen
worked at Bose Corporation, where he
held a variety of positions over the
course of 16 years. His previous titles
include: regional business manager,
Northeast territory manager, national
accounts manager, and new products
448
SOUND
TRACK
development. While working for Bose,
Breen won several awards including
Highest Attainment to Quota, Most
Improved Region, Highest Total Sales,
Largest Single Sale, and The Presi-
dent’s Award.
ABOUT COMPANIES…
Solid State Logic, an AES sustaining
member, received the prestigious
2004 Technical Grammy Award at
the recent 46th Annual Grammy
Awards ceremony on February 8, in
Los Angeles, CA. The award recog-
nizes SSL for outstanding technical
achievement in the design and pro-
duction of audio mixing consoles.
Since the company’s founding in
1969, SSL has pioneered advances in
console technology to aid the audio
production process.
Solid State Logic is a leading manu-
facturer of analog and digital audio
consoles for broadcast, postproduc-
tion, music, and film facilities.
CONFERENCES, MEETINGS…
Music China, International Exhibition
for Musical Instruments and Services,
will take place October 20-23, 2004, at
the Shanghai New International Expo
Centre. Last year, the exhibition
attracted over 21,000 trade buyers and
visitors from 49 countries and regions.
This year, the exhibition will be held
in conjunction with Prolight+Sound
Shanghai and will include seminars,
live shows, and demonstrations on the
latest trends in the musical instrument
and accessory business. Organizers of
this year’s exhibition are Messe Frank-
furt (HK) Ltd., INTEX Shanghai Co.
Ltd., and the China Music Instrument
Association (CMIA). For information
on Music China, contact:
Ms. Judy Cheung, Messe Frankfurt
(HK) Ltd.; tel: 852-2238-9933,
fax: 852-2598-8771, Internet:
www.musikmesse.com or e-mail:
judy.cheung@hongkong.messefrank-
furt.com.
Noise-Con 2004 will be held July
12-14 at the Wyndham Inner Harbor
Hotel in Baltimore, Maryland. This
joint meeting of the Institute of Noise
Control Engineering of the USA
(INCE/USA) and the Transportation-
Related Noise and Vibration Commit-
tee of the Transportation Research
Board (TRB-A1F04) will feature tech-
nical sessions on all aspects of noise
control engineering with an emphasis
on transportation noise, and an exposi-
tion of measurement instrumentation
and noise and vibration control prod-
ucts. There will also be receptions and
a dinner cruise of the Baltimore Harbor.
The conference proceedings will be
published on a CD-ROM and will be
included as part of the package each
attendee receives at the conference. In
addition to general sessions, special
sessions are being organized in the fol-
lowing areas: transportation noise,
analysis and measurements, vendor
products, and policies and markets.
For more information on INCE,
the conference and on abstract
submission, visit: www.inceusa.org.
The call for papers can also be found
on the INCE Web site: www.
inceusa.org/NoiseCon04call.pdf.
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April

AES SUSTAINING MEMBER
BROADCAST LINE MICRO-
PHONE is the BCM 104 for radio
announcing. The microphone features a
K 104 large-diaphragm condenser with
cardioid directional pattern and switch-
able proximity effect compensation. A
high-pass filter reduces frequencies
below 100 Hz by 12 dB/octave. A sec-
ond, pre-attenuation switch allows the
sensitivity to be reduced by 14 dB to
optimize performance for circuits
designed for dynamic microphones.
Both switches are internally mounted
within the microphone housing. The
BCM 104 head grille twists off easily
for quick cleaning. Additionally, a fit-
ted elastic mount prevents structure-
borne noise and is compatible with
standard broadcast-segment micro-
phone arms. Neumann, 1 Enterprise
Drive, Old Lyme, CT 06371, USA; tel.
+1 860 434 5220; fax +1 860 434
3148; Web site ww.neumannusa.com.
ACTIVE SUBWOOFER SYSTEM
is designed for large-scale installations
in either stereo or surround. The
7073A features four 12-in drivers,
fast-acting low-distortion amplifiers,
124-dB SPL capability down to 19 Hz
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April
NEW PRODUCTS
AND
DEVELOPMENTS
Product information is provided as a
service to our readers. Contact manu-
facturers directly for additional infor-
mation and please refer to the Journal
of the Audio Engineering Society.
and a 6.1-capable bass management
system for flexibility. All electronics
are integrated into the 7073A sub-
woofer cabinet, including active
crossover filters, driver overload pro-
tection circuits and power amplifiers.
The bass management system features
six inputs and outputs, LFE input and
summed signal output connectors.
Dedicated LFE input has a low-pass
filter selectable to 85 or 120 Hz, plus a
+10 dB sensitivity switch. If the LFE
channel includes frequencies above
the crossover frequency, the system’s
redirect function sends them to the
center front output to ensure they are
audible. Genelec Inc., 7 Tech Circle,
Natick, MA 01760, USA; tel. +1 508
652 0900; fax +1 508 652 0909; e-
mail genelec.usa@genelec.com; Web
site www.genelec.com.
REFERENCE/MIXING MONITOR
incorporates the drivers from the high-
ly acclaimed A-20 in a radically new
cabinet design that disperses sound
over a wider listening area. The new
M-20 features a flat, wide baffle across
the tweeter and a narrower baffle sur-
rounding the woofer for more even dis-
persion and enhanced imaging. Unlike
the A-20, there are no parallel surfaces
inside the M-20’s cabinet, so internal
standing waves remain random and
provide a frequency response of 45 Hz
to 20 kHz +/- 5 dB. Cabinet dimen-
sions are 15-in x 13-in x 10-in and
each monitor weighs 18 lbs. The M-80
is sold by the channel and includes a
250-W monaural control amplifier that
fits in a single rack space. NHTPro,
6400 Goodyear Road, Benicia, CA
94510, USA; tel. 800-648-9993 (toll
free); fax +1 707 747 1273; Web site
www.nhtpro.com.
AES SUSTAINING MEMBER
WIRELESS SYSTEM for theater and
event production is composed of the
Artist Elite 5000 System, AT899 Sub-
miniature lavalier, and a transmitter
system. The Artist Elite 5000 Wireless
System offers frequency-agile opera-
tion with IntelliScan™ frequency
selection for a choice of two hundred
selectable UHF channels. The system
also features PC-compatible control
software, easy setup and maximum
operation flexibility. The AT899cW
subminiature condenser microphone
measures 5 mm in diameter and fea-
tures a low-profile housing with inter-
nal construction designed to minimize
noise from handling, clothing, cos-
tumes, and wind. The AEW-5111 UHF
Wireless Dual-UniPak™ Transmitter
System includes an AEW-R5200 dual
receiver and two AEW-T1000 Uni-
Pak™ transmitters. Finally, at the heart
of the system is the AEW-R5200
receiver, which combines two indepen-
dent receivers in a single full-rack
housing. The 5000 Series Wireless
Systems are available with a choice of
two condenser and two dynamic hand-
held Artist Elite microphone/transmit-
ters. Audio Technica U.S., Inc., ➥
449
 NEW PRODUCTS
AND
DEVELOPMENTS
1221 Commerce Drive, Stow, OH
44224, USA; tel. +1 330 686 2600; fax
+1 330 688 3752; Web site www.
audio-technica.com.
SOFTWARE OPTION complements
the signal analyzers of the Rohde &
Schwarz FSQ series as well as the
company’s FSU and FSP spectrum
analyzers. The software expands the
applications by measurement functions
in accordance with the 3GPP2 specifi-
cations for mobile phones. Thus, mea-
surements in line with the 1 x EV-DV
standard are now feasible for the first
time. The regular cdma2000 standard,
which is supported particularly in Asia
and North America, is also covered.
Revision C of the 1xEV-DV standard
enables a higher data transmission rate
compared to the regular cdma2000
standard, allowing additional data ser-
vices. Used in combination with soft-
ware option FS-K83, the signal and
spectrum analyzers are ideally suited
for use in the development, production,
and quality assurance of mobile
phones that function in accordance
with the cdma2000 and 1xEV-DV
standards. Rohde & Schwarz GmbH &
Co. KG, Mühldorfstr. 15, D-81671
Munich, Germany; tel. +49 89 4129
13779; fax +49 89 4129 13777; e-mail
c u s t o m e r s u p p o r t @r s d . r o h d e -
schwarz.com; Web site www.rohde-
schwarz.com.
AES SUSTAINING MEMBER
ANALOG AND DIGITAL CON-
VERTERS manage complex titles for
high-resolution/multichannel music
mastering. Meitner’s ADC8 provides
accuracy in converting analog masters
to the digital domain with the proper
amount of warmth. The DAC8 sends
audio from the digital to the analog
arena with no loss of quality in the
conversion process. During the master-
ing process, the user is able to control
all of the audio sources using the Meit-
ner Switchman source controller,
which allows the user to A/B audio
from the mastering console, analog
tape decks, and DVD player. SADiE,
Inc., 475 Craighead Street, Nashville,
TN 37204, USA; tel. +1 615 327 1140;
fax +1 615 327 1699; e-mail:
sales @ sadieus.com; Web site
www.sadie.com.
450
AVAILABLE
LITERATURE
The opinions expressed are those of
the individual reviewers and are not
necessarily endorsed by the Editors of
the Journal.
CATALOGS, BROCHURES…
A new catalog of miniature micro-
phones and accessories is now avail-
able. The 34-page, full color catalog has
details of DPA’s MSS6000 Microphone
Summation System and the MPS6000
Microphone Power Supply, as well as
all current miniature microphone prod-
ucts and solutions. Also featured is the
company’s range of 33 connection
adapters, which support more than 80
different wireless transmitter systems.
A wide selection of mounts, grids,
windscreens, clips and magnets,
including the popular MHS6001 Hold-
er for Strings, and the EMK and
FMK4071 microphone kits developed
for film and broadcast use are includ-
ed. The illustrated catalog contains
specifications as well as a detailed
description of product applications.
For a copy contact: DPA Microphones
A/S, Gydevang 42-44, DK-3450
Alleroed, Denmark; tel: +45-4814-
2828, fax: +45-4814-2700, Internet:
www.dpamicrophones.com; e-mail:
info@dpamicrohpones.com.
IN BRIEF AND OF INTEREST…
Thomson ISI has announced that it is
expanding its Web of Science® cov-
erage with a Century of Science™, a
new initiative that will provide the
research community with access to the
world’s most influential scientific
research throughout the 20th Century.
This comprehensive archive, Web
of Science, The Century of Science,
includes bibliographic data from the
highest impact scientific literature
published between 1944-2000. It will
add nearly 850 000 articles from
approximately 200 journals. The
Thomson ISI editorial team carefully
selected journals based on criteria
such as citation patterns, geographic
origin and meaningful balance across
scientific disciplines. The Century of
Science initiative will extend through
2004, with new material available in
2005.
Thomson ISI, The Thomson Corpo-
ration is in Stamford, Connecticut,
USA; tel: 215-366-0100, ext. 1396,
Internet: www.thomsonisi.com.
Soundspace-Architecture for Sound
and Vision by Peter Grueneisen is
published by Birkhauser Publishing
Ltd., Basel, Switzerland. The 240-page
book is written and compiled by studio
bau:ton’s founder and principal archi-
tect. It contains issues central to build-
ing for the creation of sound, picture
and contemporary media. With over
700 illustrations, the book covers
diverse topics aimed at professionals,
students, and music and film lovers as
well as architecture aficionados.
The relationship between space,
sound, and vision is explored. Recent
developments in the music, film, and
media industries have resulted in new
types of buildings. These projects
require a comprehensive approach
from various disciplines to bridge
architecture, art and technology. The
book is based on the experiences of
studio bau:ton over 13 years in com-
bining these disciplines.
Essays and projects by renowned
experts in audio engineering, science,
music and architecture are included. A
chapter on buildings and projects cov-
ers over forty projects, which include
architectural competitions for muse-
ums and concert halls, music studios,
film facilities, TV, and radio broadcast
studios. Many of the buildings are pre-
sented in a beautifully illustrated port-
folio. The format is accessible and can
be enjoyed by beginners as well as
professionals.
The hardcover book costs $77 and is
available online at www.birkhauser.ch.
ISBN: 3-7643-6975-2.
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April
 MEMBERSHIP
INFORMATION
Section symbols are: Aachen Student Section (AA), Adelaide (ADE), Alberta (AB), All-Russian State Institute of Cinematography
(ARSIC), American River College (ARC), American University (AMU), Appalachian State University (ASU), Argentina (RA),
Atlanta (AT), Austrian (AU), Ball State University (BSU), Belarus (BLS), Belgian (BEL), Belmont University (BU), Berklee
College of Music (BCM), Berlin Student (BNS), Bosnia-Herzegovina (BA), Boston (BOS), Brazil (BZ), Brigham Young University
(BYU), Brisbane (BRI), British (BR), Bulgarian (BG), Cal Poly San Luis Obispo State University (CPSLO), California State
University–Chico (CSU), Carnegie Mellon University (CMU), Central German (CG), Central Indiana (CI), Chicago (CH), Chile
(RCH), Cincinnati (CIN), Citrus College (CTC), Cogswell Polytechnical College (CPC), Colombia (COL), Colorado (CO),
Columbia College (CC), Conservatoire de Paris Student (CPS), Conservatory of Recording Arts and Sciences (CRAS), Croatian
(HR), Croatian Student (HRS), Czech (CR), Czech Republic Student (CRS), Danish (DA), Danish Student (DAS), Darmstadt
(DMS), Del Bosque University (DBU), Detmold Student (DS), Detroit (DET), District of Columbia (DC), Duquesne University
(DU), Düsseldorf (DF), Ecuador (ECU), Expression Center for New Media (ECNM), Finnish (FIN), Fredonia (FRE), French
(FR), Full Sail Real World Education (FS), Graz (GZ), Greek (GR), Hampton University (HPTU), Hong Kong (HK), Hungarian
(HU), I.A.V.Q. (IAVQ), Ilmenau (IM), India (IND), Institute of Audio Research (IAR), Israel (IS), Italian (IT), Italian Student
(ITS), Japan (JA), Javeriana University (JU), Kansas City (KC), Korea (RK), Lithuanian (LT), Long Beach/Student (LB/S), Los
Andes University (LAU), Los Angeles (LA), Louis Lumière (LL), Malaysia (MY), McGill University (MGU), Melbourne (MEL),
Mexican (MEX), Michigan Technological University (MTU), Middle Tennessee State University (MTSU), Moscow (MOS), Music
Tech (MT), Nashville (NA), Nebraska (NEB), Netherlands (NE), Netherlands Student (NES), New Orleans (NO), New York (NY),
New York University (NYU), North German (NG), Norwegian (NOR), Ohio University (OU), Orson Welles Institute (OWI),
Pacific Northwest (PNW), Peabody Institute of Johns Hopkins University (PI), Pennsylvania State University (PSU), Peru (PER),
Philadelphia (PHIL), Philippines (RP), Polish (POL), Portland (POR), Portugal (PT), Ridgewater College, Hutchinson Campus
(RC), Romanian (ROM), Russian Academy of Music, Moscow (RAM/S), SAE Nashville (SAENA), St. Louis (STL), St. Petersburg
(STP), St. Petersburg Student (STPS), San Buenaventura University (SBU), San Diego (SD), San Diego State University (SDSU),
San Francisco (SF), San Francisco State University (SFU), Serbia and Montenegro (SAM), Singapore (SGP), Slovakian Republic
(SR), Slovenian (SL), South German (SG), Spanish (SPA), Stanford University (SU), Swedish (SWE), Swiss (SWI), Sydney (SYD),
Taller de Arte Sonoro, Caracas (TAS), Technical University of Gdansk (TUG), Texas State University—San Marcos (TSU), The
Art Institute of Seattle (TAIS), Toronto (TOR), Turkey (TR), Ukrainian (UKR), University of Arkansas at Pine Bluff (UAPB),
University of Cincinnati (UC), University of Colorado at Denver (UCDEN), University of Hartford (UH), University of Illinois at
Urbana-Champaign (UIUC), University of Luleå-Piteå (ULP), University of Massachusetts–Lowell (UL), University of Miami
(UOM), University of Michigan (UMICH), University of North Carolina at Asheville (UNCA), University of Southern California
(USC), Upper Midwest (UMW), Uruguay (ROU), Utah (UT), Vancouver (BC), Vancouver Student (BCS), Venezuela (VEN),
Vienna (VI), Webster University (WEB), West Michigan (WM), William Paterson University (WPU), Worcester Polytechnic
Institute (WPI), Wroclaw University of Technology (WUT).

These listings represent new membership according to grade.

MEMBERS Roy Markowitz Patrick Narron
800 West End Ave. Ste. 12A, New York 38439 Oliver Way, Fremont, CA 94536 (SF)
Eric Liemley City, NY 10025 (NY)
3013 Findley Rd., Kensington, MD 20895 Bryan Neumeister
(DC) Raynald Masse 2226 N. 17th Ave., Phoenix, AZ 85007
Jeff Lipton 124 de l’Emeraude, St. Colomban, J5K 2C8,
Toshiyuki Nishiguchi
Peerless Mastering & Consulting, 1085 Quebec City, Canada (MGU)
NHK Science & Technical Research Labs.,
Commonwealth Ave. #322, Boston, MA Devin Maxwell Kinuta 1-10-11, Setagaya-ku, Tokyo 157-
02215 (BOS) 36 S. 4th St. #D9, Brooklyn, NY 11211 (NY) 8510, Japan (JA)
Wing Wah Lo Thomas P. McGinty Akira Nishimura
Rm. 1001 Chung Hwa Bldg., Ma Hang 800 Starbuck Ave. Ste. A103, Watertown, Tokyo Univeristy Information Sciences,
Chung Rd., Tokwawan, Kowloon, Hong NY 13601 (NY) Yatocho 1200-2, Wakaba-ku, Chiba-shi 265-
Kong (HK) 8501, Japan(JA)
Chris Meade
Brian Loudenslager
117 Dobbin St., Brooklyn, NY 11222 (NY) Marc Nutter
102 Eddieton Ct., Mebane, NC 27302 (AT)
275 S. Gilpin St., Denver, CO 80210 (CO)
Yan-Chen Lu Michael Modesto
6F No. 192 Lien-cheng Rd., Chung-ho, Tapei 2220 Ashwood, Nashville, TN 37212 (NA) Christopher O’Malley
County, Taiwan 12535 Pacific Ave. #2, Los Angeles, CA
Masayuki Morimoto
90066 (LA)
Ian Macbeth Kobe University Faculty of Engineering,
Dulverton House, Cedar Ave., Alsager, Rokkoudai-cho 1, Nada-ku, Kobe-shi, Richard Perry
Cheshire, ST7 2PH, UK (BR) Hyougo-shi 657-8501, Japan (JA) 200 Stoneacre Court, Peachtree City, GA
30269 (AT)
Gabriel Maga James Mullen
#201-3846 Carrigan Crt., Burnaby, V3N 44 Randolph Ave., Old Bridge, NJ 08857 Andrew Peters
4H9, British Columbia, Canada (BC) (NY) P. O. Box 36411, Tucson, AZ 85740 ➥
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April 451
 MEMBERSHIP
INFORMATION
Gregory Ramos Jr. William Dave Scott Evans
215 Stagg Walk #2/B, Brooklyn, NY 11206 320 Honeylocust Ct., Bel Air, MD 21015 3003 Snelling Ave. North, St. Paul, MN
(NY) (DC) 55113 (UMW)
Stephen Rice Justin Davis Richard L. Faint
105 Lakeway Dr., Oxford, MS 38655 3351 S. Winston St., Aurora, CO 80013 (CO) 21 Redgum Ct., Bellbird Park, QLD 4300,
Todd A. Rosso Moska Davis Australia (BRI)
1090 Ashland Ave., St. Paul, MN 55104 212 E. Fulton St., Long Beach, NY 11561 James J. Falco
(UMW) (NY) 69-10B 188St. # 2B, Fresh Meadows, NY
Paul Saari Lawrence de Martin 11365 (NY)
P. O. Box 31115, Oakland, CA 94604 (SF) 2525 Arapahoe Ave. Ste. E4-272, Boulder, Nick Farina
Andres Schmidt CO 80302 (CO) 617 So. Lytle #3, Chicago, IL 60607 (CH)
2188 Tulip Rd., San Jose, CA 95128 (SF) Dan Dean
James Farina II
Yury Spitsyn 5 Lindley Rd., Mercer Island, WA 98040
3030 Flo Lor Dr. # 12, Youngstown, OH
Dartmouth College, 6242 Hallgarten Hall, (PNW)
44511 (DET)
Hanover, NH 03755 (BOS) Albertus C. den Brinker
Philips Research, WO 02, Prof. Holstlaan 4, Roberto A. Felarca
Tim Stambaugh
NL 5656 AA, Eindhoven, Netherlands (NE) 51 Zorra St. SFDM, Quezon City, MM 1105,
400 Belleville St., New Orleans, LA 70114
Philippines (RP)
(NO) Douglas Devitt
Paul Stevens 95 Newton St., Weston, MA 02493 (BOS) Dominique Fellot
188 Baddow Rd., Chelmsford, Essex CM2 6 bis rue Michel-Ange, FR 92160, Antony,
Jose Diaz-Betancourt France (FR)
9QW, UK (BR) 24 Street # J-10, Royal Town, Bayamon, PR
Albert T. Stokes Alvin Finnerty
David Dintenfass
940 Wadworth Blvd., Lakewood, CO 80214 P. O. Box 36, 104 Harrison St., Boon, MI
7549-27th Ave. NW, Seattle, WA 98117
(CO) 49618 (CH)
(PNW)
Makoto Tanase Daniel Dion Greg Fischer
Tokai Television Broadcasting Co. Ltd., 230 Rte. 195, St-Rene de Matane, G0J 3E0, 2550 So. 36th St., Quincy, IL 62305 (CH)
Higashi sakura 1-14-27, Nagoyo-shi, Aichi- Quebec City, Canada (MGU) Noel Flatt
ken, Japan (JA) 69 Jensen Rd., Watertown, MA 02472 (BOS)
Stacey Dockery
Ronald Thomas 3640 Eden Dr., Dallas, TX 75287
840 N. Hunter Dr. #4, W. Hollywood, CA Joel Foner
90069 (LA) James Dodd 4 Townsend Circle, Natick, MA 01760
260 Tower Rd., Marietta, GA 30060 (AT) (BOS)
Jeremy Todd
100 Memorial Dr. Apt. 8-15C, Cambridge, Andy Dolph Lorenzo Fontanesi
MA 02142 (BOS) 2A Waldron Ct., Dover, NH 03820 (BOS) via del Mercato 1, IT 42100, Reggio Emilia,
Thomas Domajnke Italy (IT)
Masaaki Tomita
Kitano 21, Tokororpzawa-shi, Saitama-ken 7400 N. U.S. Hwy. 1 #105, Cocoa, FL 32927 Philip Forchelli
359-1152, Japan (JA) Eric Dubowsky 859 So. Cottonwood Rd., Walnutport, PA
17 John St. #11D, New York, NY 10038 18088 (PHIL)
Thanh Tran
9707 Palacios Court, Houston, TX 77064 (NY) Tom Fowlston
Matthew Duda 315 N. Beckley Ave. , DeSoto, TX 75115
Kiyohito Uno
Muraakiyama 3-16-2, Izumi-Ku Sendai-shi, 41463 Crabtreet Ln., Plymouth, MI 48170 Jacob Freeman
Miyagi-ken 981-3205, Japan (JA) (DET) 55 E. 10th St. #506E, New York, NY 10003
Hiroshi Uto Mike Duff (NYU)
Hanazono 2-8-15 #101, Hanamigawa-ku, 2 Hively, Eureka Springs, AR 72632
Martin Frey
Chiba-shi, Chiba-ken 262-0025, Japan (JA) Steve Dupaix 500 Bonerwood Dr., Nashville, TN 37211
John M. Wallace IV P. O. Box 9660, Salt Lake City, UT 8410 (NA)
6615 English Hills Dr. 1B, Charlotte, NC (UT)
Terris Friedman
28212 (AT) Roderick Durham 415 E. 37th St., New York, NY 10016 (NY)
Terri Winston 21640 Frame Square, Ashburn, VA 20148
(DC) Olivier Fruttero
499 Alabama St. #108, San Francisco, CA P. O. Box 4283, Sabanitas, Colon, Panama
94110 (SF) Walter Dvorak
Simon Woollard Habichergasse 32, AT 1160, Vienna, Austria Brian Frye
14 Grosvenor Gardens, St. Neots, (AU) 4915 Ansley Ln., Cumming, GA 30040 (AT)
Cambridgeshire, PE19 1DL, UK (BR) H. Emerson Dwellingham John Frye
2411 Pleasant Hollow, Plainsboro, NJ 08536 3020 Willow Wisp Way, Cumming, GA
ASSOCIATES (NY) 30040 (AT)
Hermano N. Dantas Jr. Jonathan Erickson Miwa Fukino
Rua Gararu 726, Getulio Vargas, Aracaju, 4316 Sycamore, Dallas, TX 75204 Panasonic
Sergipe 49055300, Brazil (BZ) Andre Espindola 6F Kyobashi National Bldg., 2-13-10
Susan Darney Rua Tome de Souza 1234/1802, Belo Kyobashi, chuo-ku, Tokyo 104-0031, Japan
102 Thresher Ct., Cary, NC 27513 (AT) Horizonte, MG 30140131, Brazil (BZ) (JA)

452 J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April

 MEMBERSHIP
INFORMATION
Eric Gaalaas Pete Greg
18 Hartford St., Bedford, MA 01730 (BOS) AmeriCom Inc., 308 Lafayette Freeway, St.
Paul, MN 55107 UMW)
Jonathan Gardner
612 N. White Oak Ave., Midwest City, OK Gary Griffey
73130 1045 Homer Scott Rd., Bethpage, TN 37022
(NA)
Mauricio Gargel
Rua Oratorio 522 ap 601, Sao Paulo, SP Thomas Grub
03116-010, Brazil (BZ) 5 Brixton St., Flemington, Victoria 3031,
Australia (SYD)
Henri Gauci
6607 106A Ave., Edmonton, T6A 1K2, Chad Gumeringer
Alberta, Canada (AB) 418 Weatherby Way, Bismark, ND 58501
(UMW)
Hagai Gefen
6265 Variel Ave., Woodland Hills, CA Suha Gur
91367 (LA) 726 Nathan Hale Ave., Laurenceville, NJ
08648 (NY)
Otfried Geffert
Geffert Digital Engineering, Osterholder Str. Srikanth Gurrapu
17, DE 25482, Appen, Germany 5353 Keller Springs Rd. #923, Dallas, TX
75248
Stanley George
Rodney Gurule
P. O. Box 118773, Carrollton, TX 75011
5620 Blue Pine Rd. NW, Albuquerque, NM
Andreas J. Gerrits 87120
Philips Research, WO-02, Prof. Holstlaan 4,
Jari Hagqvist
NL 5656 AA, Eindhoven, Netherlands (NE)
Nokia Visiokatu 1, FI 33720, Tampere,
Joe Giardiello Finland (FIN)
61 Downing Dr., Wyomissing, PA 19610 Garrett Haines
(PHIL) 937 William Penn Ct., Pittsburgh, PA 15221
Allen Gibbs (PHIL)
P. O. Box 250064, Plano, TX 75025 Steve Hancock
Brad Gibbs 8215 Heather Ln., Mebane, NC 27302 (AT)
1500 Washington St. # 9J, Hoboken, NJ Zach Harper
07030 (NY) 7473 Harrow Dr., Nashville, TN 37221 (NA)
Enric Gine Guix Christopher Hastings
c/Arago 291 1er 2ona, ES 08009, Barcelona, 3783 Redwood Ave., Los Angeles, CA
Spain (SPA) 90066 (LA)
Leroy Gipson
P. O. Box 15414, Atlanta, GA 30333 (AT)
Karl Heilbron
5825 SW 8th St. 2nd Flr., Miami, FL 33144
Advertiser
Mark Gladden David Henderson
Signature Sight and Sound, P. O. Box 23567,
Charlotte, NC 28227 (AT)
Tonesoup Productions, 57 Cheltenham Way,
Avon, CT 06001 (NY) Internet
Andrew Goeppner David Hermann
8600 Frazier Dr., Plain City, OH 43064
(DET)
90 Wyndham St. S. # 909, Guelph, N1E 7H7,
Ontario, Canada (TOR) Directory
Ana Gonzalez Miguel Hernandez
C/Ramon Baudet Grandy 2, 3C, ES 38004, New Art Digital, Miguel Serrano 25A, Col
Santa Cruz de Tenerife, Spain (SPA) Del Valle, D.F. 01000, Mexico (MEX)
*Prism Media Products, Inc. ..............415
Rodrigo A. Gonzalez Osorio Stephan Herzog www.prismsound.com
Vargas Buston 566, San Miguel, Santiago, Friedrichstrasse 4, DE 67655, Kaiserslautern,
Chile (RCH) Germany *Neutrik, AG ........................................441
Christian Hessler www.neutrik.com
Johanne Goyette
3116 W. 112th Court, Unit F, Westminster,
9 Plac Cambrai, Outremont, H2V 1X4, *That Corporation...............................453
CO 80031 (CO)
Quebec, Canada (MGU) www.thatcorp.com
Rob Hislop
Charles Greene 101 W 23rd St. #2314, New York, NY 10011
4915 Broadway #3N, New York, NY 10034 (NY)
(NY)
Tim Hochstedler
Dale Greg 3203 SE Woodstock #1517, Portland, OR
AmeriCom Inc., 308 Lafayette Freeway, St. 97202 (POR)
Paul, MN 55107 (UMW)
Volker Hochwald
Jim Greg Alpine Electronics Europe GmbH, Vor dem *AES Sustaining Member.
AmeriCom Inc., 308 Lafayette Freeway, St. Lauch 14, DE 70567, Stuttgart, Germany
Paul, MN 55107 (UMW) (SG) ➥
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April 453
 MEMBERSHIP
INFORMATION
Auke E. Hoekstra Insung Jung Roman E. Klun
Paets van Troostwijkstraat 135-A, NL 2522, 8674 Pinyon St., Buena Park, CA 90620 38 King Street E., P. O. Box 66600, Stoney
DP S Gravenhage, Netherlands (NE) (LA) Creek, L8G 5E6, Ontario, Canada (TOR)
David Horrocks Mark Kalinowski Boney Knox Lenox
340-19th St. NE, Calgary T2E 4W9, Alberta, 205 Jefferson Ave., Bristol, PA 19007 96 Vanderveer St., Brooklyn, NY 11207
Canada (AB) (PHIL) (NY)
Bill Houston Yogesh Kamat Mhamai Christopher Konovaliv
624 Georges Ln., Ardmore, PA 19003 Impulsesoft Pvt. Ltd. 690, 15th Cross llnd 1719 Jackson Ave. #1, Ann Arbor, MI 48103
(PHIL) Phase JP Nagar, Bangalore, Karnataka (DET)
Gerard Hranek 560078, India (IND) Danny Kopelson
18318 NE 146th Way, Woodinville, WA Hafthor S. Karisson 65 W. 112th St. #3H, New York, NY 10025
98072 (PNW) Karfavogur 54, IS 104, Reykjavik, Iceland (NY)
Austin Hu Brent Karley Rena N. Kozak
6337 Odessa Dr., West Bloomfield, MI 7117 Bending Oak Rd., Austin, TX 78735 2511 39th St. NE, Calgary, T1Y 3K9,
48324 (DET) Alberta, Canada (AB)
Georgios Karpouzas
Darin Hughes James Kramer
Delfwin 56, GR 54638, Thessaloniki, Greece
2625 Lafayette Ave., Winter Park, FL 32789 16265E. Balsam Dr., Fountain Hills, AZ
(GR)
85268
Seraphim Huling Pitsch Karrer
15 Violet Circle, Milford, MA 01757 (BOS) 108 Park Terrace East #4F, New York, NY Allan Kraut
10034 (NY) 25 Parkdale Ave., Boston, MA 02134 (BOS)
Jason Hurley
887 Oak Valley Ln., Nashville, TN 37220 Satoh Kazue Timothy P. Krips
(NA) Neya 676-73, Neyagawa-shi, Osaka-fu 572- 2413 Beach Rd., Walled Lake, MI 48390
0801, Japan (JA) (DET)
Aaron Hurst
157 Center Square Rd., Leola, PA 17540 Lew Kellogg V. Kumareswaran
(PHIL) 2008 Petworth Court, Raleigh, NC 27615 SAE Malaysia, No.10-1 Jalan USJ 9/5R,
(AT) Subang Business Centre, Subang Jaya,
Robert Iadanza Selangor 47620, Malaysia (MY)
4 Carvel Rd., Monroe, NY 10950 (NY) Antti P. Kelloniemi
Helsinki University of Technology, P. O. Box Toshifumi Kunimoto
Giovanni Imlach 2-28-17 Jouhoku, Hamamatsu-shi, Shizuoka
1000, FI 02015, TKK / Espoo, Finland (FIN)
227 E. 96th St. #2FE, New York, NY 10128 432-8011, Japan (JA)
(NY) Jason Kemmerer
646 Benson Way, Thousand Oaks, CA 91360 Chee Meng Kwan
Morgan Inman (LA) No.7 Jalan Nikmat, Happy Garden, Jalan
Man Made Music, 16 W. 46th St. 6th Flr., Kuchai Lama, Kuala Lumpur 58200,
New York, NY 10036 (NY) Gregory J. Kephart
11334 Niland Pass, Ft. Wayne, IN 46845 Malaysia (MY)
Kazuo Izumoya (CH) Geronimo Labrada
Takaido-higashi, 3-14-17, Suginami-ku, Exequiel Fernandez 1624 dpto 62-A,
Tokyo 168-0072, Japan (JA) James Keyes
194 S. Buckhout St., Irvington, NY 10533 Santiago, Chile (RCH)
Jerry Jackman (NY) Rodney Ladson
1061 E 1100 N., Orem, UT 84097 (UT) 4807 51st Pl., Hyattsville, MD 20781 (DC)
Gary Khan
Dean Jakubowicz 1231 W. Barry, Chicago, IL 60657 (CH) Rene Laflamme
AmeriCom Inc., 308 Lafayette Freeway, St. Youngtae Kim 9276 La Jeunesse St. 5, Montreal, H2M 1S2,
Paul, MN 55107 (UMW) SAIT, P. O. Box 111, Suwon, Korea (RK) Quebec, Canada, (MGU)
Eric Jensen Sung-Jun Kim Gary Lamb
20-53 19th St. #2A, Astoria, NY 11105 (NY) Northern Caribbean University, Music Baderstrasse 4, CH 8400, Winterthur,
Jin Soo. Jeon Department, Mandeville, Jamaica Switzerland (SWI)
203 Edgeland St., Rochester, NY 14609 Renzo Larrauri Cunza
Patrick Kimball
(NY) Hondurad 301, Lima 18, Peru (PER)
2738 Winston Rd., Oklahoma City, OK
Samuel Johnson 73120 Drew Lavyne
33 Pine Grove, Amherst, MA 01102 (BOS) 9 E. 13th St., Suite 5I, New York, NY 10003
Joshua King (NY)
Marty Johnson 8780 N. Mount Tabor Rd., Ellettsville, IN
143 Durham Hall, Dept. Mech. Eng. Virginia 47429 (CI) Sergio R. Ledesma
Tech., Blacksburg, VA 24061 (DC) Arquitectos 26-3, Colonia Escandon, D.F.
Charles King 01800, Mexico (MEX)
Eric Johnson 11 Oakwood Ln., Greenwich, CT 06830
1410 Midtown Place, Raleigh, NC 27609 (BOS) Hyuck J. Lee
Samsung Electronics Co. Ltd., 416 Maetan 3
Eric Johnson David Kingland - dong, Paldal-2u, Suwon, Korea (RK)
3401 J. St. # 3, Sacramento, CA 95816 (SF) 103 Shady Oaks Ln., Lake Mills, IA 50450
Michael Lee
Echo Jong (UMW)
3132 David Ave., Palo Alto, CA 94303 (SF)
23 Hsin Hua Rd., Taoyuan 330, Taiwan Bjarni B. Kjartansson David Lee
Larry Josephson Hringbraut 37, IS 107, Reykjavik, Iceland Concept Associates Sdn Bhd, 44 Jalan
Radio Foundation Inc., 1 W. 89th St., New Gregory Klas Sungkai, off Jalan Terengganu, Georgetown,
York, NY 10024 (NY) 8820 Sunset Dr., Colden, NY 14033 (NY) Penang 10460, Malaysia (MY)

454 J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April

 MEMBERSHIP
INFORMATION
Jamie Lendino Marshall Simmons Anne-Marie Sylvestre
128 Bay 13th St., Brooklyn, NY 11214 (NY) 224 Varsity N., Bowling Green, OH 43402 6251 de Normanville, Montreal, H2S 2B7
Rich Leone (UC) Quebec, Canada (MGU)
12810 5th St. N, Lake Elmo, MN 55042 Steven Simpson Dennes K. Szilagyi
(UMW) 4837 Perina Way, North Highlands, CA Dirk Vreekenstraat 53, NL 1019 DP,
95660 (ARC) Amsterdam, Netherlands (NES)
Robert Lessick
Randy Tan
178 Dougbeth Dr., Ronkonkoma, NY 11779 Jasmina Sironja
23 Kent Pl. #6, Menlo Park, CA 94025 (SFU)
(NY) Bul. Umetnosti 11, YU 11070, Novi
Beograd, Yugoslavia Anton Teljeback
Opal Leung Korsgatan 13 A, SE 903 36, Umea, Sweden
123 Williams St., Jamaica Plain, MA 02130 Andy Skies Nenad Terentic
(BOS) 5102 W. Piedmont, Laveen, AZ 85339 Ralja, YU 11311, Smederevo B.B.,
Chin-Ping Liao (CRAS) Yugoslavia
P. O. Box 10-35 Nei Hou, Taipei 114, Keith Smith Nicolas Thelliez
Taiwan Naval Media Center, PSC 4 66 Box 14, FPO 146 avenue de Flandre, FR 75019, Paris,
Heung-Guy Lim AP 96595, Virgin Islands France (CPS)
881 Constitution Dr., Foster City, CA 94404 Ann Smith Katrina Thomas
(SF) 8210 Rambleton Way, Antelope, CA 95843 3629 Sweet Grass Circle #6031, Winter Park,
Yat Fung Jim Yam (ARC) FL 32792 (FS)
Flat D 33/F Tower 2 New Haven, Hong Cameron Snow Michael Thompson
Kong (HK) 50 Mill Ln., Huntington, NY 11743 (IAR) 1639 Birch St., State College, PA 16801
Geof Lipman (PSU)
Frederic Soulard
425 Prospect Place Unit 5H, Brooklyn, NY 27 avenue Secretan, FR 75019, Paris, France Anastasia Timofeyeva
11238 (NY) (CPS) 5425 Kohler Rd., Sacramento, CA 95841
Jeff Lipton (ARC)
Austin Sousa
Peerless Mastering, 1085 Commonwealth 2205 2nd Ave. #605, Seattle, WA 98121 Donald E. Tonelli
Ave # 322, Boston, MA 02215 (BOS) (TAIS) 8040 Sunrise Blvd. #A, Citrus Heights, CA
Ed Littman 95610 (ARC)
Esther L. Spence
250 W 10th St., New York, NY 10014 (NY) 6696 W. Long Dr., Littleton, CO 80123 Milan Trifunovic
David Logvin (UCDEN) Kumanovska 28/1, YU 11420, Smederevska
21 Glen Ave., Chelmsford, MA 01824 (BOS) Palanka, Yugoslavia
Joshua Srago
Richard Long 1600 Holloway Ave., El Cerrito, CA 94132 Brandon Trowel
2904 Austin Ave., Waco, TX 76710 (USF) 7313 Holworthy Way, Sacramento, CA
95842 (ARC)
Robert Long Jason J. Steeber
1583 Semoran North Circle # 203, Winter 1810 Bauman Rd., Quakertown, PA 18951 Miki Tsutsumi
Park, FL 32792 (PSU) 30-10 28th Rd. #2, Astoria, NY 11102 (IAR)
Jason Love Ise K. Steffensen Ryan M. Tucker
10719 Green Valley Rd., Union Bridge, MD Oernevej 4, 3 Fl., DK 2400 NV, Copenhagen, 1402 Country Club Rd. #38, Norfolk, NE
21791 (DC) Denmark (DAS) 68701

John Lucas James Vajda

Aaron T. Stefl
2 Princeton Ct., Tinton Falls, NJ 07724 (NY) 168 Mass. Ave., SB# 2929, Boston, MA
7541 Quail Nest Pl., Citrus Heights, CA
02115 (BCM)
Ricciano Lumpkins 95610 (ARC)
Platinum Work Projects LLC, P. O. Box Alex Valdes
Kevin Steller 5121 N. 40th St. #B-214, Phoenix, AZ 85018
948182, Maitland, FL 32794 6 Park Pl., Newburgh, NY 12550 (FRE) (CRAS)
Jennifer Lyons Maya Stevens Dan Valente
411 Camino del Rio South, Suite 301, San 564 Price St., Daly City, CA 94014 (CPC) 1033 Danby Rd. #1-02, Ithaca, NY 14850
Diego, CA 92108 (SD)
Daniel Stover (FRE)
Martin F. Macheiner 229 I-45 N. #102, Conroe, TX 77304 Elisa Valentini
Gartengasse 45, AT 2721, Bad Fischau -
Gerda Strobl via delle Cornacchie 927/01, IT 55100, San
Brunn, Austria (AU)
Lessingstrasse 12/1, AT 8010, Graz, Austria Marco - Lucca, Italy (ITS)
Anthony Magrath (GZ) Daniel Van Ampting
Flat 4, 52 Bath St., Edinburgh EH15 1HF,
Charles Stumbagh 901 DeKleva Dr., Apopka, FL 32712 (FS)
UK (BR)
222 SE Main #5, Mesa, AZ 85213 (CRAS) Raymond van Weeghel
Dominick Maita
Shiva Sundaram Klinkenberg 4, NL 8091 GV, Wezep,
235 E. 22nd St. 7L, New York, NY 10010
Signal and Image Processing Institute, Netherlands (NES)
(NY)
University of Southern California, EEB400, Jeroen M. van Zaalen
STUDENTS 340 Mclintock Ave., Los Angeles, CA 90089 Evert de Bruijnstraat 108, NL 1411 TE,
(USC) Naarden, Netherlands (NES)
Jeff Siarto
821 Bay Pointe Dr., Oxford, MI Bobby Sweeney Vincent Verderame
48371(MTU) 702 Douglas, Stanton, NE 68779 57 Springcress Dr., Delran, NJ 08075 (WPU) ➥

J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April 455

 MEMBERSHIP
INFORMATION In Memoriam
Lisa G. Vitti
145 Bay Ridge Parkway #2, Brooklyn, NY
11209 (IAR)

G
Barbara Vlahides len Akins, AES life member, accomplished with speed, innova-
301 E. 22nd St., #11D, New York, NY 10010 died of heart failure on tion, and economy.
(IAR) November 14, 2003, in Los Akins retired in 1977 after leading
Aaron Vlasnik Angeles, CA. He was 87 years old. his departments with a calm, wise,
801 E. Benjamin, Dorm Box 218, Norfolk, Atkins was born in York, PA, in and humanistic statesmanship that
NE 68701
1916. While attending Gettysburg belied the frenzy of television broad-
Jelle Vlietstra College, he worked as a theater pro- casting. He then traveled the U. S.
Appelhof 7, NL 9201 KT, Drachten, jectionist, which spurred his interest in and worldwide with his wife Alice,
Netherlands (NES)
all things audio. Trained as an elec- visiting all continents but Antarctica.
Robin Watkins tronics engineer, Glen was employed Audio was equally important as
467 Napoleon Ave., Columbus, OH 43213 by the International Telephone and video in Glen’s mind. He encouraged
(HPTU)
Telegraph in their short wave and promoted the design and con-
Bethany Watson radio transmitter final test department. struction of production audio con-
2827 N. Spaulding #3, Chicago, IL 60618 In 1942, he joined the Office of soles and communication systems,
(CC)
War Information, and after a series which were not available commer-
Felipe Wernet of harrowing transportation incidents cially at the time. At home, he was an
Schrijnwerker 12, NL 3201 TK, Spijkenisse,
spent four years in China setting up avid gardener, an ardent amateur
Netherlands (NES)
and maintaining radio transmitting radio operator, and a hi-fi audio
Daniel White equipment, which often involved enthusiast. Loudspeaker systems
1135 S. 15th #4, Lincoln, NE 68502
dubious political consequences. This were his particular interest. He built
Felton White position involved associations with much of his own equipment. Ham
P.O. Box 5275, River Forest, IL 60305 (CC) prominent journalists and political swap meets and conventions contin-
Dennis Wikstroem figures in the area, which led to an ued to be a part of his life. He never
Ankarskatavagen 83 A, SE 941 31, Pitea, additional post of gathering and stopped learning.
Sweden (ULP) broadcasting news from China for Don McCroskey
Andreas Woerle the CBS radio network.
Zimmerplatzgasse 1, AT 8010, Graz, Austria In 1948, Glen joined the sound oseph Habig, AES life mem-
(GZ)
Endale Worku
1831 8th Ave. #502, Seattle, WA 98101
(TAIS)
department of RKO Studios in Holly-
wood, CA. When commercial televi-
sion became a reality, he moved to
J ber, died of Parkinson’s disease
on September 21, 2003, in Tin-
ton Falls, NJ. He was 79 years old.
ABC television, Hollywood, in 1951. Born in New York City, Habig
Christopher Wraith Hired as a video projectionist, he received a bachelor’s degree in music
The Coach House, 46 Painshawfield Rd., rapidly advanced to positions involv- education at the City College of New
Stocksfield, Northumberland, NE43 7QY,
UK ing the planning and maintenance of York. He also attended Julliard and
new equipment and facilities. He the Manhattan School of Music.
Fan Wu developed a working 3-D video sys- Trained as a classical musician, he
P. O. Box 147, Beijing Broadcasting Institute,
Beijing 100024, Peoples Republic of China tem, which was demonstrated at the worked with many of the top classi-
1953 NAB convention. cal artists during his career as a
Phillip Yarrow
In 1960, he designed the first producer. He also recorded major
750 Font Blvd. #C332, San Francisco, CA
94132 (SFU) video/audio/machine control popular and jazz artists.
routing switcher that served 24 A Grammy Award-winning record
Chris Yates
297 Commssioners Rd. E., London, N6C
sources to six control rooms. producer, Habig won the award for
2T3, Ontario, Canada Automation of video and audio his recording of Stravinsky’s Sym-
switching for the local station phony of Psalms. For 10 years he
Anthony N. Yeager
813 Sullivan Dr., Lansdale, PA 19446 control room was another pro- played the trombone with various
ject that involved unique symphony orchestras. He later
Emi S. Yonemura approaches. Many landmark became an artists’ and repertoire pro-
5312 Markwood Ln., Fair Oaks, CA 95628
(ARC) broadcasts bore his fingerprints, ducer at RCA Victor Red Seal for 19
such as the 1960 Democratic years (1954 to 1973). After that he
Luis G. Zamora Manriquez
convention, Nixon Kennedy de- worked as an executive producer at
Avenida Chillan 27 85, Independencia,
Santiago, Chile bates, and the Wide World of Reader’s Digest Music Division for
Sports. The conversion to color 16 years. He is survived by his wife,
Jose R. Zapata Gonzalez
cra 64A #39-15 Apto. 102, Medellin,
television under his aegis was Virginia Harlow Habig.
Antioquia, Colombia (LAU)

456 J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April

 AUDIO ENGINEERING SOCIETY 26th Conference
October 1–3

CALL for PAPERS

AES 26TH CONFERENCE, 2004
High-Quality Analog Audio Processing
Baarn, The Netherlands
Dates: October 1–3, 2004, Location: Baarn, The Netherlands
Chair: Ben Kok, Dorsserblesgraaf, NL, Email: 26th_chair@aes.org

The AES 26th International Conference intends to explore the new insights in analog audio technology that have contributed to the
overall increase in the subjective and objective quality of modern digital audio systems. The resolution of digital audio systems,
both in the time domain and in the amplitude domain, has undergone a spectacular improvement in recent years. Because nearly
all digital audio signals are derived from analog microphone signals, the recording industry has directed new efforts to the design
of low-level and line-level analog circuitry to keep up with the increasing demands of the digital audio world. In particular, analog
microphone amplifiers and associated circuits such as line drivers, power supplies, cables, etc., have to match the quality of the
modern high-resolution A-to-D converters. Recent years have also seen an increase in the attention paid to a system’s ability to
create an illusion of depth, of space surrounding the performers, and of the feeling of “being there.” The relationship of these sub-
jective experiences with aspects of the design of the equipment will be one of the main topics of this conference. Because of the
subjective nature of this field, a preference will be given to papers that combine a lecture with a listening demonstration. For these
demonstrations three identical first-class listening rooms equipped for stereo listening will be available at the Polyhymnia Studios.
The AES 26th Conference Committee invites submission of technical papers and proposals for demonstrations at the conference in
October 2004 in Baarn. By 2004 May 17, a proposed title, 60- to 120-word abstract, and 500- to 1000-word precis of the paper
should be submitted via the Internet to the AES26th Conference paper-submission site at www.aes.org/26th_authors. You can
visit this site for more information and complete instructions for using the site anytime after 2004 March 19. The author’s informa-
tion, title, abstract, and precis should all be submitted online. The precis should describe the work performed, methods employed,
conclusion(s), and significance of the paper. Titles and abstracts should follow the guidelines in Information for Authors at
www.aes.org/journal/con_infoauth.html. Acceptance of papers will be determined by the 26th Conference review committee
based on an assessment of the abstract and precis.

PROPOSED TOPICS FOR PAPERS

New Circuits and Topologies EMC Considerations
Feedback: Bane or Blessing Passive Components
Power Supplies and Audible Effects Tubes and Transistors
Discrete versus Integrated Circuits Connectors and Contacts
Filters, Bandwidth Effects Transformers
Correlation Between Measurements and Cables
Listening Experiences Lay-out and Mechanical Design
SUBMISSION OF PAPERS SCHEDULE
Please submit proposed title, abstract, and precis Proposal deadline: 2004 May 17
at www.aes.org/26th_authors no later than 2004 Acceptance emailed: 2004 June 9
May 17. If you have any questions, contact:
Paper deadline: 2004 July 19
PAPERS COCHAIRS
Tom Magchielse Peter van Willenswaard Authors whose contributions have been
Consultant Audiomagic accepted for presentation will receive
The Netherlands The Netherlands additional instructions for submission
Email: 26th_papers@aes.org of their manuscripts.

J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April 457


The following is the latest information we have available for our sections contacts. If you
wish to change the listing for your section, please mail, fax or e-mail the new information
to: Mary Ellen Ilich, AES Publications Office, Audio Engineering Society, Inc., 60 East
42nd Street, Suite 2520, New York, NY 10165-2520, USA. Telephone +1 212 661 8528,
ext. 23. Fax +1 212 661 7829. E-mail MEI@aes.org.
Updated information that is received by the first of the month will be published in the
next month’s Journal. Please help us to keep this information accurate and timely.
EASTERN REGION,
USA/CANADA
Vice President:
Jim Anderson
12 Garfield Place
Brooklyn, NY 11215
Tel. +1 718 369 7633
Fax +1 718 669 7631
E-mail vp_eastern_usa@aes.org
UNITED STATES OF
AMERICA
CONNECTICUT
University of Hartford
Section (Student)
Timothy Britt
Faculty Advisor
AES Student Section
University of Hartford
Ward College of Technology
200 Bloomfield Ave.
West Hartford, CT 06117
Tel. +1 860 768 5358
Fax +1 860 768 5074
E-mail aes@hartfordaes.org
FLORIDA
Full Sail Real World
Education Section (Student)
Bill Smith, Faculty Advisor
AES Student Section
Full Sail Real World Education
3300 University Blvd., Suite 160
Winter Park, FL 327922
Tel. +1 800 679 0100
E-mail full_sail@aes.org
University of Miami Section
(Student)
Ken Pohlmann, Faculty Advisor
AES Student Section
University of Miami
School of Music
PO Box 248165
Coral Gables, FL 33124-7610
Tel. +1 305 284 6252
Fax +1 305 284 4448
E-mail miami@aes.org
GEORGIA
Atlanta Section
Robert Mason
2712 Leslie Dr.
Atlanta, GA 30345
Tel./Fax +1 770 908 1833
E-mail atlanta@aes.org
MARYLAND
Peabody Institute of Johns
Hopkins University Section
(Student)
Neil Shade, Faculty Advisor
458
SECTIONS CONTACTS
DIRECTORY
AES Student Section
Peabody Institute of Johns
Hopkins University
Recording Arts & Science Dept.
2nd Floor Conservatory Bldg.
1 E. Mount Vernon Place
Baltimore, MD 21202
Tel. +1 410 659 8100 ext. 1226
E-mail peabody@aes.org
MASSACHUSETTS
Berklee College of Music
Section (Student)
Eric Reuter, Faculty Advisor
Berklee College of Music
Audio Engineering Society
c/o Student Activities
1140 Boylston St., Box 82
Boston, MA 02215
Tel. +1 617 747 8251
Fax +1 617 747 2179
E-mail berklee@aes.org
Boston Section
J. Nelson Chadderdon
c/o Oceanwave Consulting, Inc.
21 Old Town Rd.
Beverly, MA 01915
Tel. +1 978 232 9535 x201
Fax +1 978 232 9537
E-mail boston@aes.org
University of Massachusetts
–Lowell Section (Student)
John Shirley, Faculty Advisor
AES Student Chapter
University of Massachusetts–Lowell
Dept. of Music
35 Wilder St., Ste. 3
Lowell, MA 01854-3083
Tel. +1 978 934 3886
Fax +1 978 934 3034
E-mail umass_lowell@aes.org
Worcester Polytechnic
Institute Section (Student)
William Michalson
Faculty Advisor
AES Student Section
Worcester Polytechnic Institute
100 Institute Rd.
Worcester, MA 01609
Tel. +1 508 831 5766
E-mail wpi@aes.org
NEW JERSEY
William Paterson University
Section (Student)
David Kerzner, Faculty Advisor
AES Student Section
William Paterson University
300 Pompton Rd.
Wayne, NJ 07470-2103
Tel. +1 973 720 3198
Fax +1 973 720 2217
E-mail wpu@aes.org
NEW YORK
Fredonia Section (Student)
Bernd Gottinger, Faculty Advisor
AES Student Section
SUNY–Fredonia
1146 Mason Hall
Fredonia, NY 14063
Tel. +1 716 673 4634
Fax +1 716 673 3154
E-mail fredonia@aes.org
Institute of Audio Research
Section (Student)
Noel Smith, Faculty Advisor
AES Student Section
Institute of Audio Research
64 University Pl.
New York, NY 10003
Tel. +1 212 677 7580
Fax +1 212 677 6549
E-mail iar@aes.org
New York Section
Bill Siegmund
Digital Island Studios
71 West 23rd Street Suite 504
New York, NY 10010
Tel. +1 212 243 9753
E-mail new_york@aes.org
New York University Section
(Student)
Robert Rowe, Faculty Advisor
Steinhardt School of Education
35 West 4th St., 777G
New York, NY 10012
Tel. +1 212 998 5435
E-mail nyu@aes.org
NORTH CAROLINA
Appalachian State University
Section (Student)
Michael S. Fleming
Faculty Advisor
Sonaura Sound
152 Villafe Drive
Boone, NC 28607
Tel. +1 828 263 0454
E-mail appalachian@aes.org
University of North Carolina
at Asheville Section (Student)
Wayne J. Kirby
Faculty Advisor
AES Student Section
University of North Carolina at
Asheville
Dept. of Music
One University Heights
Asheville, NC 28804
Tel. +1 828 251 6487
Fax +1 828 253 4573
E-mail north_carolina@aes.org
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April
PENNSYLVANIA
Carnegie Mellon University
Section (Student)
Thomas Sullivan
Faculty Advisor
AES Student Section
Carnegie Mellon University
University Center Box 122
Pittsburg, PA 15213
Tel. +1 412 268 3351
E-mail carnegie_mellon@aes.org
Duquesne University Section
(Student)
Francisco Rodriguez
Faculty Advisor
AES Student Section
Duquesne University
School of Music
600 Forbes Ave.
Pittsburgh, PA 15282
Tel. +1 412 434 1630
Fax +1 412 396 5479
E-mail duquesne@aes.org
Pennsylvania State University
Section (Student)
Dan Valente
AES Penn State Student Chapter
Graduate Program in Acoustics
217 Applied Science Bldg.
University Park, PA 16802
Home Tel. +1 814 863 8282
Fax +1 814 865 3119
E-mail penn_state@aes.org
Philadelphia Section
Rebecca Mercuri
P. O. Box 1166.
Philadelphia, PA 19105
Tel. +1 215 327 7105
E-mail philly@aes.org
VIRGINIA
Hampton University Section
(Student)
Bob Ransom, Faculty Advisor
AES Student Section
Hampton University
Dept. of Music
Hampton, VA 23668
Office Tel. +1 757 727 5658,
+1 757 727 5404
Home Tel. +1 757 826 0092
Fax +1 757 727 5084
E-mail hampton_u@aes.org
WASHINGTON, DC
American University Section
(Student)
Rebecca Stone-gordon
Faculty Advisor
AES Student Section
American University
4400 Massachusetts Ave., N.W.
 SECTIONS CONTACTS
DIRECTORY
Washington, DC 20016 Faculty Advisor
Tel. +1 202 885 3242 AES Student Section
E-mail american_u@aes.org 676 N. LaSalle, Ste. 300
Chicago, IL 60610
District of Columbia Section Tel. +1 312 344 7802
John W. Reiser
Fax +1 312 482 9083
DC AES Section Secretary E-mail columbia@aes.org
P. O. Box 169
University of Illinois at
Mt. Vernon, VA 22121-0169
Tel. +1 703 780 4824 Urbana-Champaign Section
Fax +1 703 780 4214 (Student)
E-mail dc@aes.org Mark Hasegawa-Johnson,
Faculty Advisor
CANADA AES Student Section
University of Illinois, Urbana-
McGill University Section Champaign
(Student) Urbana, IL 61801
John Klepko, Faculty Advisor E-mail urbana@aes.org
AES Student Section
McGill University INDIANA
Sound Recording Studios
Ball State University Section
Strathcona Music Bldg. (Student)
555 Sherbrooke St. W.
Michael Pounds, Faculty Advisor
Montreal, Quebec H3A 1E3 AES Student Section
Canada
Ball State University
Tel. +1 514 398 4535 ext. 0454 MET Studios
E-mail mcgill_u@aes.org
2520 W. Bethel
Toronto Section Muncie, IN 47306
Anne Reynolds Tel. +1 765 285 5537
606-50 Cosburn Ave. Fax +1 765 285 8768
Toronto, Ontario M4K 2G8 E-mail ball_state@aes.org
Canada
Tel. +1 416 957 6204 Central Indiana Section
James Latta
Fax +1 416 364 1310
E-mail toronto@aes.org Sound Around
6349 Warren Ln.
Brownsburg, IN 46112
CENTRAL REGION, Office Tel. +1 317 852 8379
F Ridgewater College, Hutchinson
USA/CANADA Fax +1 317 858 8105
E-mail central_indiana@aes.org
Vice President: KANSAS
Frank Wells
Kansas City Section
2130 Creekwalk Drive Jim Mitchell
Murfreesboro, TN
Custom Distribution Limited
Tel. +1 615 848 1769 12301 Riggs Rd.
Fax +1 615 848 1108 Overland Park, KS 66209
E-mail vp_central_usa@aes.org Tel. +1 913 661 0131
Fax +1 913 663 5662
UNITED STATES OF
AMERICA LOUISIANA
New Orleans Section
ARKANSAS Joseph Doherty
University of Arkansas at Factory Masters
Pine Bluff Section (Student) 4611 Magazine St.
Robert Elliott, Faculty Advisor New Orleans, LA 70115
AES Student Section Tel. +1 504 891 4424
Music Dept. Univ. of Arkansas Cell +1 504 669 4571
at Pine Bluff Fax +1 504 899 9262
1200 N. University Drive E-mail jdoherty@accesscom.net
Pine Bluff, AR 71601
Tel. +1 870 575 8916 MICHIGAN
Fax +1 870 543 8108 Detroit Section
E-mail pinebluff@aes.org David Carlstrom
DaimlerChrysler
ILLINOIS E-mail detroit@aes.org
Chicago Section
Tom Miller Michigan Technological
Knowles Electronics University Section (Student)
1151 Maplewood Dr. Greg Piper
AES Student Section
Itasca, IL 60143
Tel. +1 630 285 5882 Michigan Technological
University
Fax +1 630 250 0575
E-mail chicago@aes.org 1400 Townsend Dr.
121 EERC Building
Columbia College Section Houghton, MI 49931
(Student) Tel. +1 906 482 3581
Dominique J. Chéenne E-mail michigan_tech@aes.org
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April
University of Michigan
Section (Student)
Jason Corey,
Faculty Advisor
University of Michigan School
of Music
1100 Baits Drive
Ann Arbor, MI 48109
E-mail univ_michigan@aes.org
West Michigan Section
Carl Hordyk
Calvin College
3201 Burton S.E.
Grand Rapids, MI 49546
Tel. +1 616 957 6279
Fax +1 616 957 6469
E-mail west_mich@aes.org
MINNESOTA
Music Tech College Section
(Student)
Michael McKern
Faculty Advisor
AES Student Section
Music Tech College
19 Exchange Street East
Saint Paul, MN 55101
Tel. +1 651 291 0177
Fax +1 651 291 0366
E-mail
musictech_student@aes.org
Ridgewater College,
Hutchinson Campus Section
(Student)
Dave Igl, Faculty Advisor
AES Student Section
Campus
2 Century Ave. S.E.
Hutchinson, MN 55350
E-mail ridgewater@aes.org
Upper Midwest Section
Greg Reierson
Rare Form Mastering
4624 34th Avenue South
Minneapolis, MN 55406
Tel. +1 612 327 8750
E-mail upper_midwest@aes.org
MISSOURI
St. Louis Section
John Nolan, Jr.
693 Green Forest Dr.
Fenton, MO 63026
Tel./Fax +1 636 343 4765
E-mail st_louis@aes.org
Webster University Section
(Student)
Faculty Advisor:
Gary Gottleib
Webster University
470 E. Lockwood Ave.
Webster Groves, MO 63119
Tel. +1 961 2660 x7962
E-mail webster_st_louis@aes.org
NEBRASKA
Nebraska Section
Anthony D. Beardslee
Northeast Community College
P.O. Box 469
Norfolk, NE 68702
Tel. +1 402 844 7365
Fax +1 209 254 8282
E-mail nebraska@aes.org
OHIO
Cincinnati Section
Dan Scherbarth
Digital Groove Productions
5392 Conifer Dr.
Mason, OH 45040
Tel. +1 513 325 5329
E-mail cincinnati@aes.org
Ohio University Section
(Student)
Erin M. Dawes
AES Student Section
Ohio University, RTVC Bldg.
9 S. College St.
Athens, OH 45701-2979
Home Tel. +1 740 597 6608
E-mail ohio@aes.org
University of Cincinnati
Section (Student)
Thomas A. Haines
Faculty Advisor
AES Student Section
University of Cincinnati
College-Conservatory of Music
M.L. 0003
Cincinnati, OH 45221
Tel. +1 513 556 9497
Fax +1 513 556 0202
E-mail Cincinnati@aes.org
TENNESSEE
Belmont University Section
(Student)
Wesley Bulla, Faculty Advisor
AES Student Section
Belmont University
Nashville, TN 37212
E-mail Belmont@aes.org
Middle Tennessee State
University Section (Student)
Phil Shullo, Faculty Advisor
AES Student Section
Middle Tennessee State University
301 E. Main St., Box 21
Murfreesboro, TN 37132
Tel. +1 615 898 2553
E-mail mtsu@aes.org
Nashville Section
Tom Edwards
MTV Networks
330 Commerce St.
Nashville, TN 37201
Tel. +1 615 335 8520
Fax +1 615 335 8625
E-mail nashville@aes.org
SAE Nashville Section (Student)
Larry Sterling, Faculty Advisor
AES Student Section
7 Music Circle N.
Nashville, TN 37203
Tel. +1 615 244 5848
Fax +1 615 244 3192
E-mail saenash_student@aes.org
TEXAS
Texas State University—San
Marcos (Student)
Mark C. Erickson
Faculty Advisor
AES Student Section
Southwest Texas State University
224 N. Guadalupe St.
San Marcos, TX 78666
Tel. +1 512 245 8451
Fax +1 512 396 1169
E-mail tsu_sm@aes.org
459
 SECTIONS CONTACTS
DIRECTORY
WESTERN REGION,
USA/CANADA
Vice President:
Bob Moses
Island Digital Media Group,
LLC
26510 Vashon Highway S.W.
Vashon, WA 98070
Tel. +1 206 463 6667
Fax +1 810 454 5349
E-mail vp_western_usa@aes.org
UNITED STATES OF
AMERICA
ARIZONA
Conservatory of The
Recording Arts and Sciences
Section (Student)
Glenn O’Hara, Faculty Advisor
AES Student Section
Conservatory of The Recording
Arts and Sciences
2300 E. Broadway Rd.
Tempe, AZ 85282
Tel. +1 480 858 9400, 800 562
6383 (toll-free)
Fax +1 480 829 1332
E-mail
conservatory_RAS@aes.org
CALIFORNIA
American River College
Section (Student)
Eric Chun, Faculty Advisor
AES Student Section
American River College Chapter
4700 College Oak Dr.
Sacramento, CA 95841
Tel. +1 916 484 8420
E-mail american_river@aes.org
Cal Poly San Luis Obispo
State University Section
(Student)
Bryan J. Mealy
Faculty Advisor
AES Student Section
California Polytechnic State
University
Dept. of Electrical Engineering
San Luis Obispo, CA 93407
Tel. +1 805 756 2300
Fax +1 805 756 1458
E-mail san_luis_obispo@aes.org
California State University
–Chico Section (Student)
Keith Seppanen, Faculty Advisor
AES Student Section
California State University–Chico
400 W. 1st St.
Chico, CA 95929-0805
Tel. +1 530 898 5500
E-mail chico@aes.org
Citrus College Section (Student)
Stephen O’Hara, Faculty Advisor
AES Student Section
Citrus College
Recording Arts
1000 W. Foothill Blvd.
Glendora, CA 91741-1899
Fax +1 626 852 8063
Cogswells Polytechnical
College Section (Student)
Tim Duncan, Faculty Sponsor
460
AES Student Section
Cogswell Polytechnical College
Music Engineering Technology
1175 Bordeaux Dr.
Sunnyvale, CA 94089
Tel. +1 408 541 0100, ext. 130
Fax +1 408 747 0764
E-mail cogswell@aes.org
Expression Center for New
Media Section (Student)
Scott Theakston, Faculty Advisor
AES Student Section
Ex’pression Center for New Media
6601 Shellmount St.
Emeryville, CA 94608
Tel. +1 510 654 2934
Fax +1 510 658 3414
E-mail expression_center@aes.org
Long Beach City College
Section (Student)
Nancy Allen, Faculty Advisor
AES Student Section
Long Beach City College
4901 E. Carson St.
Long Beach, CA 90808
Tel. +1 562 938 4312
Fax +1 562 938 4409
E-mail long_beach@aes.org
Los Angeles Section
Andrew Turner
2311 Lakeview Ave.
Los Angeles, CA 90039
Tel. +1 323 663 1327
E-mail la@aes.org
San Diego Section
J. Russell Lemon
2031 Ladera Ct.
Carlsbad, CA 92009-8521
Home Tel. +1 760 753 2949
E-mail san_diego@aes.org
San Diego State University
Section (Student)
John Kennedy, Faculty Advisor
AES Student Section
San Diego State University
Electrical & Computer
Engineering Dept.
5500 Campanile Dr.
San Diego, CA 92182-1309
Tel. +1 619 594 1053
Fax +1 619 594 2654
E-mail sdsu@aes.org
San Francisco Section
Conrad Cooke
1046 Nilda Ave.
Mountain View, CA 94040
Office Tel. +1 650 846 1132
Home Tel. +1 650 321 0713
E-mail san_francisco@aes.org
San Francisco State
University Section (Student)
John Barsotti, Faculty Advisor
AES Student Section
San Francisco State University
Broadcast and Electronic
Communication Arts Dept.
1600 Halloway Ave.
San Francisco, CA 94132
Tel. +1 415 338 1507
E-mail sfsu@aes.org
Stanford University Section
(Student)
Jay Kadis, Faculty Advisor
Stanford AES Student Section
Stanford University
CCRMA/Dept. of Music
Stanford, CA 94305-8180
Tel. +1 650 723 4971
Fax +1 650 723 8468
E-mail stanford@aes.org
University of Southern
California Section (Student)
Kenneth Lopez
Faculty Advisor
AES Student Section
University of Southern California
840 W. 34th St.
Los Angeles, CA 90089-0851
Tel. +1 213 740 3224
Fax +1 213 740 3217
E-mail usc@aes.org
COLORADO
Colorado Section
Roy Pritts
2873 So. Vaughn Way
Aurora, CO 80014
Tel. +1 303 369 9514
E-mail colorado@aes.org
University of Colorado at
Denver Section (Student)
Roy Pritts, Faculty Advisor
AES Student Section
University of Colorado at Denver
Dept. of Professional Studies
Campus Box 162
P.O. Box 173364
Denver, CO 80217-3364
Tel. +1 303 556 2795
Fax +1 303 556 2335
E-mail cu_denver@aes.org
OREGON
PORTLAND SECTION
Tony Dal Molin
Audio Precision, Inc.
5750 S.W. Arctic Dr.
Portland, OR 97005
Tel. +1 503 627 0832
Fax +1 503 641 8906
E-mail portland@aes.org
UTAH
Brigham Young University
Section (Student)
Timothy Leishman,
Faculty Advisor
BYU-AES Student Section
Department of Physics and
Astronomy
Brigham Young University
Provo, UT 84602
Tel. +1 801 422 4612
E-mail brigham_young@aes.org
Utah Section
Deward Timothy
c/o Poll Sound
4026 S. Main
Salt Lake City, UT 84107
Tel. +1 801 261 2500
Fax +1 801 262 7379
WASHINGTON
Pacific Northwest Section
Gary Louie
University of Washington
School of Music
P. O. Box 353450
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April
Seattle, WA 98195
Office Tel. +1 206 543 1218
Fax +1 206 685 9499
E-mail pacific_nw@aes.org
The Art Institute of Seattle
Section (Student)
David G. Christensen
Faculty Advisor
AES Student Section
The Art Institute of Seattle
2323 Elliott Ave.
Seattle, WA 98121-1622
Tel. +1 206 448 0900
E-mail
art_institute_seattle@aes.org
CANADA
Alberta Section
Frank Lockwood
AES Alberta Section
Suite 404
815 - 50 Avenue S.W.
Calgary, Alberta T2S 1H8
Canada
Home Tel. +1 403 703 5277
Fax +1 403 762 6665
E-mail alberta@aes.org
Vancouver Section
David Linder
93.7 JRfm/600am Radio, A
Division of the Jim Pattison
Broadcast Group
E-mail vancouver@aes.org
Vancouver Student Section
Gregg Gorrie, Faculty Advisor
AES Greater Vancouver
Student Section
Centre for Digital Imaging and
Sound
3264 Beta Ave.
Burnaby, B.C. V5G 4K4, Canada
Tel. +1 604 298 5400
E-mail
vancouver_student @ aes.org
NORTHERN REGION,
EUROPE
Vice President:
Søren Bech
Bang & Olufsen a/s
CoreTech
Peter Bangs Vej 15
DK-7600 Struer, Denmark
Tel. +45 96 84 49 62
Fax +45 97 85 59 50
E-mail
vp_northern_europe@aes.org
BELGIUM
Belgian Section
Hermann A. O. Wilms
AES Europe Region Office
Zevenbunderslaan 142, #9
BE-1190 Vorst-Brussels, Belgium
Tel. +32 2 345 7971
Fax +32 2 345 3419
DENMARK
Danish Section
Preben Kvist
c/o DELTA Acoustics &
Vibration
Bygning 356
 SECTIONS CONTACTS
DIRECTORY
Akademivej Tel. +7 095 2502161, +7 095 80-952 Gdansk, Poland Central German Section
DK 2800 Lyngby, Denmark 1929011 Tel. +48 58 347 2717 Ernst-Joachim Völker
Tel. +45 61 33 45 81 Fax +7 095 9430006 Fax +48 58 347 1114 Institut für Akustik und
Fax +45 45 20 12 01 E-mail moscow@aes.org E-mail Bauphysik
E-mail denmark@aes.org vp_central_europe@aes.org Kiesweg 22-24
Russian Academy of Music DE-61440 Oberursel, Germany
Danish Student Section Student Section AUSTRIA Tel. +49 6171 75031
Preben Kvist Igor Petrovich Veprintsev Austrian Section Fax +49 6171 85483
c/o DELTA Acoustics & Faculty Advisor Franz Lechleitner E-mail c_german@aes.org
Vibration Sound Engineering Division Lainergasse 7-19/2/1
Bygning 356 30/36 Povarskaya Street AT-1230 Vienna, Austria Darmstadt Section (Student)
Akademivej RU 121069, Moscow, Russia Office Tel. +43 1 4277 29602 G. M. Sessler, Faculty Sponsor
DK 2800 Lyngby, Denmark Tel. +7 095 291 1532 Fax +43 1 4277 9296 AES Student Section
Tel. +45 61 33 45 81 E-mail russian_academy@aes.org E-mail austria @aes.org Technical University of
Fax +45 45 20 12 01 St. Petersburg Section Darmstadt
E-mail denmark@aes.org Irina A. Aldoshina Graz Section (Student) Institut für Übertragungstechnik
St. Petersburg University of Robert Höldrich Merkstr. 25
FINLAND Faculty Sponsor DE-64283 Darmstadt, Germany
Telecommunications
Finnish Section Gangutskaya St. 16, #31 Institut für Elektronische Musik Tel. +49 6151 162869
Kalle Koivuniemi RU-191187 St. Petersburg und Akustik E-mail
Nokia Research Center Russia Inffeldgasse 10 darmstadt_student@aes.org
P.O. Box 100 Tel. +7 812 272 4405 AT-8010 Graz, Austria
FI-33721 Tampere, Finland Tel. +43 316 389 3172 Detmold Section
Fax +7 812 316 1559 (Student)
Tel. +358 7180 35452 E-mail st_petersburg@aes.org Fax +43 316 389 3171
Fax +358 7180 35897 E-mail graz_student@aes.org Andreas Meyer, Faculty Sponsor
E-mail finland@aes.org St. Petersburg Student Section AES Student Section
Natalia V. Tyurina Vienna Section (Student) c/o Erich Thienhaus Institut
NETHERLANDS Faculty Advisor Jürg Jecklin, Faculty Sponsor Tonmeisterausbildung
Prosvescheniya pr., 41, 185 Vienna Student Section Hochschule für Musik
Netherlands Section Universität für Musik und
Rinus Boone RU-194291 St. Petersburg, Russia Detmold
Tel. +7 812 595 1730 Darstellende Kunst Wien Neustadt 22, DE-32756
Voorweg 105A Institut für Elektroakustik und
NL-2715 NG Zoetermeer Fax +7 812 316 1559 Detmold, Germany
E-mail Experimentelle Musik Tel/Fax +49 5231 975639
Netherlands Rienösslgasse 12
Tel. +31 15 278 14 71, +31 62 st_petersburg_student@aes.org E-mail detmold@aes.org
AT-1040 Vienna, Austria
127 36 51 SWEDEN Tel. +43 1 587 3478 Düsseldolf Section
Fax +31 79 352 10 08 Fax +43 1 587 3478 20 (Student)
E-mail netherlands@aes.org Swedish Section
Ingemar Ohlsson E-mail vienna_student@aes.org Ludwig Kugler
Netherlands Student Section AES Student Section
Audio Data Lab CZECH REPUBLIC Bilker Allee 126
Maurik van den Steen Katarinavägen 22
AES Student Section Czech Section DE-40217 Düsseldorf, Germany
SE-116 45 Stockholm, Sweden Tel. +49 211 3 36 80 38
Prins Willemstraat 26 Jiri Ocenasek
2584 HV Den Haag, Netherlands Tel. +46 8 644 58 65 E-mail
Dejvicka 36
Tel. +31 6 45702051 Fax +46 8 641 67 91 duesseldorf_student@aes.org
CZ-160 00 Prague 6, Czech
E-mail E-mail sweden@aes.org
Republic Ilmenau Section
netherlands_student@aes.org University of Luleå-Piteå Home Tel. +420 2 24324556 (Student)
NORWAY Section (Student) E-mail czech@aes.org Karlheinz Brandenburg
Lars Hallberg, Faculty Sponsor Faculty Advisor
Norwegian Section AES Student Section Czech Republic Student AES Student Section
Jan Erik Jensen University of Luleå-Piteå Section Fraunhofer Institute for Digital
Nøklesvingen 74 School of Music Libor Husník, Faculty Advisor Media Technology IDMT
NO-0689 Oslo, Norway Box 744 AES Student Section Langewiesener Str. 22
Office Tel. +47 22 24 07 52 S-94134 Piteå, Sweden Czech Technical Univ. at Prague DE-98693 Ilmenau, Germany
Home Tel. +47 22 26 36 13 Tel. +46 911 726 27 Technická 2, Tel. +49 3677 69 4340
Fax +47 22 24 28 06 Fax +46 911 727 10 CZ-116 27 Prague 6 E-mail ilmenau_student@aes.org
E-mail norway@aes.org E-mail lulea_pitea@aes.org Czech Republic
Tel. +420 2 2435 2115 North German Section
RUSSIA UNITED KINGDOM E-mail husnik@feld.cvut.cz Reinhard O. Sahr
All-Russian State Institute of British Section Eickhopskamp 3
GERMANY
Cinematography Section Heather Lane DE-30938 Burgwedel, Germany
(Student) Audio Engineering Society Aachen Section (Student) Tel. +49 5139 4978
Leonid Sheetov, Faculty Sponsor P. O. Box 645 Michael Vorländer Fax +49 5139 5977
AES Student Section Slough SL1 8BJ Faculty Advisor E-mail n_german@aes.org
All-Russian State Institute of United Kingdom Institut für Technische Akustik
Cinematography (VGIK) Tel. +44 1628 663725 RWTH Aachen South German Section
W. Pieck St. 3 Fax +44 1628 667002 Templergraben 55 Gerhard E. Picklapp
RU-129226 Moscow, Russia E-mail uk@aes.org D-52065 Aachen, Germany Landshuter Allee 162
Tel. +7 095 181 3868 Tel. +49 241 807985 DE-80637 Munich, Germany
Fax +7 095 187 7174 Fax +49 241 8888214 Tel. +49 89 15 16 17
E-mail all_russian_state@aes.org CENTRAL REGION, E-mail aachen@aes.org Fax +49 89 157 10 31
EUROPE E-mail s_german@aes.org
Moscow Section Berlin Section (Student)
Michael Lannie Vice President: Bernhard Güttler HUNGARY
Research Institute for Bozena Kostek Zionskirchstrasse 14
Television and Radio Multimedia Systems DE-10119 Berlin, Germany Hungarian Section
Acoustic Laboratory Department Tel. +49 30 4404 72 19 István Matók
12-79 Chernomorsky bulvar Gdansk University of Technology Fax +49 30 4405 39 03 Rona u. 102. II. 10
RU-113452 Moscow, Russia Ul. Narutowicza 11/12 E-mail berlin@aes.org HU-1149 Budapest, Hungary
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April 461
 SECTIONS CONTACTS
DIRECTORY
Home Tel. +36 30 900 1802
Fax +36 1 383 24 81
E-mail hungary@aes.org
LITHUANIA
Lithuanian Section
Vytautas J. Stauskis
Vilnius Gediminas Technical
University
Traku 1/26, Room 112
LT-2001 Vilnius, Lithuania
Tel. +370 5 262 91 78
Fax +370 5 261 91 44
E-mail lithuania@aes.org
POLAND
Polish Section
Andrzej Dobrucki
Wroclaw University of
Technology
Institute of Telecommunication
and Acoustics
Wybrzeze Wyspiannkiego 27
PL-50-370 Wroclaw, Poland
Tel. +48 48 71 320 3068
Fax +48 71 320 3189
E-mail poland@aes.org
Technical University of Gdansk
Section (Student)
Pawel Zwan
AES Student Section
Technical University of Gdansk
Sound Engineering Dept.
ul. Narutowicza 11/12
PL-80 952 Gdansk, Poland
Home Tel. +48 58 347 23 98
Office Tel. +4858 3471301
Fax +48 58 3471114
E-mail gdansk_u @aes.org
Wroclaw University of
Technology Section (Student)
Andrzej B. Dobrucki
Faculty Sponsor
AES Student Section
Institute of Telecommunications
and Acoustics
Wroclaw Univ.Technology
Wybrzeze Wyspianskiego 27
PL-503 70 Wroclaw, Poland
Tel. +48 71 320 30 68
Fax +48 71 320 31 89
E-mail
wroclaw @ aes.org
REPUBLIC OF BELARUS
Belarus Section
Valery Shalatonin
Belarusian State University of
Informatics and
Radioelectronics
vul. Petrusya Brouki 6
BY-220027 Minsk
Republic of Belarus
Tel. +375 17 239 80 95
Fax +375 17 231 09 14
E-mail
belarus @ aes.org
SLOVAK REPUBLIC
Slovakian Republic Section
Richard Varkonda
Centron Slovakia Ltd.
Podhaj 107
SK-841 03 Bratislava
Slovak Republic
462
Tel. +421 7 6478 0767
Fax. +421 7 6478 0042
E-mail
slovakian_rep @aes.org
SWITZERLAND
Swiss Section
Joël Godel
AES Swiss Section
Sonnmattweg 6
CH-5000 Aarau
Tel./Fax +41 26 670 2033
Switzerland
E-mail swiss@aes.org
UKRAINE
Ukrainian Section
Dimitri Danyuk
32-38 Artyoma St., Apt. 38
UA 04053 Kiev, Ukraine
E-mail ukrainian@aes.org
SOUTHERN REGION,
EUROPE
Vice President:
Ivan Stamac
Ivlje 4
HR-10040 Zagreb, Croatia
Tel. +385 1 4822361
Home Tel./Fax +385 1 4574403
E-mail
vp_southern_europe@aes.org
BOSNIA-HERZEGOVINA
Bosnia-Herzegovina Section
Jozo Talajic
Bulevar Mese Selimovica 12
BA-71000 Sarajevo
Bosnia–Herzegovina
Tel. +387 33 455 160
Fax +387 33 455 163
E-mail
bosnia_herzegovina@aes.org
BULGARIA
Bulgarian Section
Konstantin D. Kounov
Bulgarian National Radio
Technical Dept.
4 Dragan Tzankov Blvd.
BG-1040 Sofia, Bulgaria
Tel. +359 2 65 93 37, +359 2
9336 6 01
Fax +359 2 963 1003
E-mail
bulgarian @ aes.org
CROATIA
Croatian Section
Ivan Stamac
Ivlje 4
HR-10040 Zagreb, Croatia
Tel. +385 1 4822361
Home Tel./Fax +385 1 4574403
E-mail croatian@aes.org
Croatian Student Section
Hrvoje Domitrovic
Faculty Advisor
AES Student Section
Faculty of Electrical
Engineering and Computing
Dept. of Electroaocustics (X. Fl.)
Unska 3
HR-10000 Zagreb, Croatia
Tel. +385 1 6129 640 R. Paulo Renato 1, 2A
Fax +385 1 6129 852 PT-2745-147 Linda-a-Velha
E-mail Portugal
croatian_student@aes.org Tel. +351 214145827
E-mail portugal @ aes.org
FRANCE
ROMANIA
Conservatoire de Paris
Section (Student) Romanian Section
Alessandra Galleron Marcia Taiachin
36, Ave. Parmentier Radio Romania
FR-75011 Paris, France 60-62 Grl. Berthelot St.
Tel. +33 1 43 38 15 94 RO-79756 Bucharest, Romania
E-mail Paris_student@aes.org Tel. +40 1 303 12 07
Fax +40 1 222 69 19
French Section E-mail romanian@aes.org
Michael Williams
Ile du Moulin SERBIA AND MONTENEGRO
62 bis Quai de l’Artois Serbia and Montenegro
FR-94170 Le Perreux sur Section
Marne, France Tomislav Stanojevic
Tel. +33 1 48 81 46 32 Sava centre
Fax +33 1 47 06 06 48 M. Popovica 9
E-mail french@aes.org YU-11070 Belgrade, Yugoslavia
Tel. +381 11 311 1368Fax +381
Louis Lumière Section 11 605 578
(Student) E-mail
Alexandra Carr-Brown serbia_montenegro@aes.org
AES Student Section
Ecole Nationale Supérieure SLOVENIA
Louis Lumière
7, allée du Promontoire, BP 22 Slovenian Section
Tone Seliskar
FR-93161 Noisy Le Grand
Cedex, France RTV Slovenija
Tel. +33 6 18 57 84 41 Kolodvorska 2
E-mail louis_lumiere@aes.org SI-1550 Ljubljana, Slovenia
Tel. +386 61 175 2708
Fax +386 61 175 2710
GREECE
E-mail
Greek Section slovenian @ aes.org
Vassilis Tsakiris
Crystal Audio SPAIN
Aiantos 3a Vrillissia
GR 15235 Athens, Greece Spanish Section
Juan Recio Morillas
Tel. + 30 2 10 6134767 Spanish Section
Fax + 30 2 10 6137010 C/Florencia 14 3oD
E-mail greek@aes.org ES-28850 Torrejon de Ardoz
(Madrid), Spain
ISRAEL
Tel. +34 91 540 14 03
Israel Section E-mail spanish @ aes.org
Ben Bernfeld Jr.
H. M. Acustica Ltd. TURKEY
20G/5 Mashabim St. Turkish Section
IL-45201 Hod Hasharon, Israel Sorgun Akkor
Tel./Fax +972 9 7444099 STD
E-mail israel@aes.org Gazeteciler Sitesi, Yazarlar
Sok. 19/6
ITALY Esentepe 80300
Istanbul, Turkey
Italian Section
Carlo Perretta Tel. +90 212 2889825
c/o AES Italian Section Fax +90 212 2889831
Piazza Cantore 10 E-mail turkish@aes.org
IT-20134 Milan, Italy
Tel. +39 338 9108768
Fax +39 02 58440640 LATIN AMERICAN REGION
E-mail italian@aes.org
Vice President:
Italian Student Section Mercedes Onorato
Franco Grossi, Faculty Advisor Talcahuano 141
AES Student Section Buenos Aires, Argentina
Viale San Daniele 29 Tel./Fax +5411 4 375 0116
IT-33100 Udine, Italy E-mail
Tel. +39 0432227527 vp_latin_american@aes.org
E-mail italian_student@aes.org
ARGENTINA
PORTUGAL
Argentina Section
Portugal Section German Olguin
Rui Miguel Avelans Coelho Talcahuano 141
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April
 SECTIONS CONTACTS
DIRECTORY
Buenos Aires, Argentina 1013 Tel./Fax +52 55 5240 1203 Fax +618 8 8384 3419 MALAYSIA
Tel./Fax +5411 4 375 0116 E-mail mexican @ aes.org E-mail adelaide@aes.org Malaysia Section
E-mail argentina@aes.org C. K. Ng
PERU
Brisbane Section
BRAZIL David Ringrose King Musical Industries
Orson Welles Institute Section AES Brisbane Section Sdn Bhd
Brazil Section (Student) P.O. Box 642 Lot 5, Jalan 13/2
José Carlos Giner Javier Antón Roma St. Post Office MY-46200 Kuala Lumpur
Rua Marechal Cantuária # 18 Av. Salaberry 3641, San Isidro Brisbane, Qld. AU-4003, Australia Malaysia
Urca-Rio de Janeiro Lima, Peru Office Tel. +61 7 3364 6510 Tel. +603 7956 1668
RJ-2291-060, Brazil Tel. +51 1 264 1773 E-mail brisbane@aes.org Fax +603 7955 4926
Tel. +55 21 2244 6530 Fax +51 1 264 1878 E-mail malaysia@aes.org
Fax +55 21 2244 7113 E-mail orsonwelles@aes.org Melbourne Section
E-mail brazil@aes.org Graham J. Haynes PHILIPPINES
PERU SECTION P.O. Box 5266
CHILE Wantirna South, Victoria Philippines Section
Armando Puente De La Vega AU-3152, Australia Dario (Dar) J. Quintos
Chile Section Av. Salaberry 3641 San Isidro Tel. +61 3 9887 3765 125 Regalia Park Tower
Andres Schmidt Lima, Peru Fax +61 3 9887 1688 P. Tuazon Blvd., Cubao
Hernan Cortes 2768 Tel. +51 1 264 1773 E-mail Quezon City, Philippines
Ñuñoa, Santiago de Chile Fax +51 1 264 1878 melbourne @ aes.org Tel./Fax +63 2 4211790, +63 2
Tel. +56 2 4249583 E-mail peru@aes.org 4211784
E-mail chile@aes.org Sydney Section E-mail philippines@aes.org
URUGUAY Howard Jones
COLOMBIA AES Sydney Section SINGAPORE
Uruguay Section P.O. Box 766
Colombia Section César Lamschtein Singapore Section
Sandra Carolina Hernandez Crows Nest, NSW AU-2065 Kenneth J. Delbridge
Universidad ORT Australia
CR 14 #87-25 Cuareim 1451 480B Upper East Coast Rd.
Bogotá, Colombia Tel. +61 2 9417 3200 Singapore 466518
Montevideo, Uruguay Fax +61 2 9417 3714
Tel. +57 1 622 1282 Tel. +59 1 902 1505 Tel. +65 9875 0877
Fax +57 1 629 7313 E-mail sydney@aes.org Fax +65 6220 0328
Fax +59 1 900 2952
E-mail colombia@aes.org E-mail uruguay@aes.org E-mail singapore@aes.org
HONG KONG
Javeriana University Section VENEZUELA
(Student) Hong Kong Section
Silvana Medrano Taller de Arte Sonoro, Henry Ma Chi Fai
Caracas Section (Student) HKAPA, School of Film and STUDENT DELEGATE
Carrera 7 #40-62
Bogota, Colombia Carmen Bell-Smythe de Leal Television ASSEMBLY
Tel./Fax +57 1 320 8320 Faculty Advisor 1 Gloucester Rd.
E-mail javeriana@aes.org AES Student Section Wanchai, Hong Kong
Tel. +852 2584 8824 NORTH/SOUTH
Taller de Arte Sonoro AMERICA REGIONS
Los Andes University Section Ave. Rio de Janeiro Fax +852 2588 1303
(Student) Qta. Tres Pinos E-mail Chair:
Jorge Oviedo Martinez Chuao, VE-1061 Caracas hong_kong @ aes.org Marie Desmarteau
Transversal 44 # 96-17 Venezuela McGill University Section
Bogota, Colombia Tel. +58 14 9292552 INDIA
(AES)
Tel./Fax +57 1 339 4949 ext. Tel./Fax +58 2 9937296 72 Delaware Avenue
2683 E-mail caracas@aes.org India Section
Avinash Oak Ottawa K2P 0Z3
E-mail losandes@aes.org Ontario, Canada
Venezuela Section Avisound
San Buenaventura University Elmar Leal Home Tel. +1 613 236 5411
A-20, Deepanjali Office Tel. +1 514 398 4535
Section (Student) Ave. Rio de Janeiro Shahaji Raje Marg
Nicolas Villamizar Qta. Tres Pinos E-mail
Vile Parle East tonmaestra@hotmail.com
Transversal 23 # 82-41 Apt. 703 Chuao, VE-1061 Caracas
Int.1 Venezuela Mumbai IN-400 057, India
Tel. +91 22 26827535 Vice Chair:
Bogota, Colombia Tel. +58 14 9292552 Felice Santos-Martin
Tel. +57 1 616 6593 Tel./Fax +58 2 9937296 E-mail
American River College (AES)
Fax +57 1 622 3123 E-mail venezuela@aes.org webmaster@aesindia.org Tel. +1 916 802 2084
E-mail sanbuenaventura@aes.org E-mail felicelazae@hotmail.com
JAPAN
ECUADOR INTERNATIONAL REGION
Japan Section
Ecuador Section Katsuya (Vic) Goh EUROPE/INTERNATIONAL
Juan Manuel Aguillo Vice President:
Neville Thiele 2-15-4 Tenjin-cho, Fujisawa-shi REGIONS
Av. La Prensa 4316 y Vaca de Kanagawa-ken 252-0814, Japan
Castro 10 Wycombe St.
Epping, NSW AU-2121, Tel./Fax +81 466 81 0681 Chair:
Quito, Ecuador E-mail aes_japan@aes.org Natalia Teplova
Tel./Fax +59 32 2598 889 Australia
Tel. +61 2 9876 2407 European Student Section
E-mail ecuador@aes.org KOREA Bratislavskaya Street 13-1-48
Fax +61 2 9876 2749
I.A.V.Q. Section (Student) E-mail vp_international@aes.org Moscow, RU 109 451, Russia
Korea Section Tel. +7 095 291 1532
Felipe Mardones Seong-Hoon Kang
315 Carrion y Plaza AUSTRALIA
Taejeon Health Science College Vice Chair:
Quito, Ecuador Adelaide Section Dept. of Broadcasting Martin Berggren
Tel./Fax +59 3 225 61221 David Murphy Technology European Student Section
E-mail iavq@aes.org Krix Loudspeakers 77-3 Gayang-dong Dong-gu Varvsgatan 35
MEXICO 14 Chapman Rd. Taejeon, Korea Arvika, SE 67133, Sweden
Hackham AU-5163 Tel. +82 42 630 5990 Home Tel. +46 0570 12018
Mexican Section South Australia Fax +82 42 628 1423 Office Tel. +46 0570 38500
Jorge urbano Tel. +618 8 8384 3433 E-mail korea@aes.org E-mail martin.bergren@imh.se
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April 463
AES CONVENTIONS AND CON
The latest details on the following events are posted on the AES Website: http://www.aes.org
116th Convention
Berlin, Germany
2004 Date: 2004 May 8–11
Berlin, Germany Location: Messe Berlin
Berlin, Germany
25th International
Conference
London, UK
“Metadata for Audio”
Date: 2004 June 17–19
26th International
26th Conference Conference
October 1–3 Baarn, The Netherlands
“High-Quality Analog Audio
Processing”
Date: 2004 October 1–3
117th Convention
San Francisco, CA, USA
117 th
Date: 2004 October 28–31
2004 Location: Moscone Center
San Francisco, CA, USA
San Francisco
All of the papers from AES conventions and conferences through
2003 are available on the 20-disk AES Electronic Library. The 2003
update disks for the Electronic Library are now available.
464
Convention chair: Papers cochair:
Reinhard O. Sahr Ben Bernfeld
Eickhopskamp 3
DE-30938 Burgwedel, Germany
Krozinger Str. 22
DE-79219 Staufen, Germany
Telephone: + 49 5139 4978 Email: 116th_papers@aes.org
Fax: + 49 5139 5977 Papers cochair:
Email: 116th_chair@aes.org Stephan Peus
Vice chair: Georg Neumann GmbH
Jörg Knothe Email: 116th_papers@aes.org
DeutschlandRadio
Email: 116th_vicechair@aes.org
Conference chair: Papers cochair:
John Grant Gerhard Stoll
Nine Tiles Networks, Cambridge, UK IRT, Munich, Germany
Email: 25th_chair@aes.org Email: 25th_papers@aes.org
Papers cochair:
Russell Mason
University of Surrey, Guildford, UK
Email: 25th_papers@aes.org
Conference chair: Papers cochair:
Ben Kok Tom Magchielse
Dorsserblesgraaf, The Netherlands Consultant, The Netherlands
Email: 26th_chair@aes.org Email: 26th_papers@aes.org
Papers cochair:
Peter van Willenswaard
Audiomagic, The Netherlands
Email: 26th_papers@aes.org
Convention chair: Papers cochair:
John Strawn Brian Link
S Systems
15 Willow Avenue
Dolby Laboratories
Email: 117th_papers@aes.org
Larkspur, CA, 94939 USA
Papers cochair:
Telephone: + 1 415 927 8856 Rob Maher
Fax: + 1 415 927 2935 Montana State University-Bozeman
Email: 117th_chair@aes.org Email: 117th_papers@aes.org
For price and ordering information go to www.aes.org,
send email to Andy Veloz at aav@aes.org,
or call any AES office at +1 212 661 8528, ext. 39 (USA),
+44 1628 663725 (UK), +33 1 4881 4632 (Europe).
J. Audio Eng. Soc., Vol. 52, No. 4, 2004 April
FERENCES
Exhibit information:
Thierry Bergmans
Telephone: +32 2 345 7971
Fax: +32 2 345 3419
Email: 116th_exhibits@aes.org
Call for papers: Vol. 51, No. 7/8,
pp. 768 (2003 July/August)
Convention preview, Vol. 52, No. 3,
pp. 266–287 (2004 March)
Call for papers: Vol. 51, No. 9,
pp. 871 (2003 September)
Conference preview, This issue,
pp. 402–411 (2004 April)
Call for papers: This issue,
pp. 457 (2004 April)
Exhibit information:
Chris Plunkett/Donna Vivero
Telephone: +1 212 661 8528, ext. 30
Fax: +1 212 682 0477
Email: 117th_exhibits@aes.org
Call for papers: ,Vol. 52, No. 3
pp. 319 (2004 March)
Reports of recent AES
conventions and conferences are
now available online, go to
www.aes.org/events/reports.
INFORMATION FOR AUTHORS
Presentation should be labeled with author’s name and
Manuscripts submitted should be figure number.
typewritten on one side of ISO size A4 Photographs should be black and
(210 x 297 mm) or 216-mm x 280-mm white prints without a halftone screen,
(8.5-inch x 11-inch) paper with 40-mm preferably 200 mm x 250 mm (8 inch by
(1.5-inch) margins. All copies including 10 inch).
abstract, text, references, figure captions,
Line drawings (graphs or sketches) can be
and tables should be double-spaced. original drawings on white paper, or high-
Pages should be numbered consecutively. quality photographic reproductions.
Authors should submit an original plus
two copies of text and illustrations. The size of illustrations when printed in the
Journal is usually 82 mm (3.25 inches)
Review wide, although 170 mm (6.75 inches) wide
Manuscripts are reviewed anonymously can be used if required. Letters on original
by members of the review board. After the illustrations (before reduction) must be large
reviewers’ analysis and recommendation enough so that the smallest letters are at
to the editors, the author is advised of least 1.5 mm (1/16 inch) high when the
either acceptance or rejection. On the illustrations are reduced to one of the above
basis of the reviewers’ comments, the widths. If possible, letters on all original
editor may request that the author make illustrations should be the same size.
certain revisions which will allow the
Units and Symbols
paper to be accepted for publication.
Metric units according to the System of
Content International Units (SI) should be used.
Technical articles should be informative For more details, see G. F. Montgomery,
and well organized. They should cite “Metric Review,” JAES, Vol. 32, No. 11,
original work or review previous work, pp. 890–893 (1984 Nov.) and J. G.
giving proper credit. Results of actual McKnight, “Quantities, Units, Letter
experiments or research should be Symbols, and Abbreviations,” JAES, Vol.
included. The Journal cannot accept 24, No. 1, pp. 40, 42, 44 (1976 Jan./Feb.).
unsubstantiated or commercial statements. Following are some frequently used SI
Organization units and their symbols, some non-SI units
An informative and self-contained that may be used with SI units (▲), and
some non-SI units that are deprecated (■).
abstract of about 60 words must be
provided. The manuscript should develop Unit Name Unit Symbol
the main point, beginning with an ampere A
introduction and ending with a summary bit or bits spell out
or conclusion. Illustrations must have bytes spell out
informative captions and must be referred decibel dB
to in the text. degree (plane angle) (▲) °
References should be cited numerically in farad F
brackets in order of appearance in the gauss (■) Gs
text. Footnotes should be avoided, when gram g
possible, by making parenthetical henry H
remarks in the text. hertz Hz
hour (▲) h
Mathematical symbols, abbreviations, inch (■) in
acronyms, etc., which may not be familiar joule J
to readers must be spelled out or defined kelvin K
the first time they are cited in the text. kilohertz kHz
kilohm kΩ
Subheads are appropriate and should be liter (▲) l, L
inserted where necessary. Paragraph megahertz MHz
division numbers should be of the form 0 meter m
(only for introduction), 1, 1.1, 1.1.1, 2, 2.1, microfarad µF
2.1.1, etc. micrometer µm
References should be typed on a microsecond µs
manuscript page at the end of the text in milliampere mA
order of appearance. References to millihenry mH
periodicals should include the authors’ millimeter mm
names, title of article, periodical title, millivolt mV
volume, page numbers, year and month minute (time) (▲) min
of publication. Book references should minute (plane angle) (▲) ’
contain the names of the authors, title of nanosecond ns
book, edition (if other than first), name oersted (■) Oe
and location of publisher, publication year, ohm Ω
and page numbers. References to AES pascal Pa
convention preprints should be replaced picofarad pF
with Journal publication citations if the second (time) s
preprint has been published. second (plane angle) (▲) ”
siemens S
Illustrations tesla T
Figure captions should be typed on a volt V
separate sheet following the references. watt W
Captions should be concise. All figures weber Wb
AES
sustaining
member
organizations