Professional Documents
Culture Documents
A Physical Model of A Single-Reed Wind Instrument, Including Actions of The Player
A Physical Model of A Single-Reed Wind Instrument, Including Actions of The Player
A Physical Model of A Single-Reed Wind Instrument, Including Actions of The Player
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms
The MIT Press is collaborating with JSTOR to digitize, preserve and extend access to
Computer Music Journal
This content downloaded from 154.59.124.94 on Fri, 14 Sep 2018 14:13:59 UTC
All use subject to https://about.jstor.org/terms
Eric Ducasse
Ecole Nationale Sup6rieure d'Arts et M6tiers
A Physical Model
of a Single-Reed Wind
C.E.R. de Bordeaux-Talence
33405 Talence Cedex, France
eric. ducasse@bordeaux. ensam.fr
Instrument, Including
Actions of the Player
Today, the musical interest of physical modeling presents the tone-hole model including the key or
synthesis is well defined (e.g., Keefe 1992; Smith finger motion. Each important element of the
1996). In particular, time-domain simulation of the model is illustrated by sound examples included in
physical operation of musical instruments makes the it 2003 Computer Music Journal Sound Anthol-
possible to create transitory and perceptive phe- ogy.
nomena that are difficult to obtain by other data-
processing methods. The design of the complete A Modular Representation of the Instrument
model (including the simplified modeling of the
main actions of the player) of a single-reed instru- The postulate of modularity is necessary for various
ment presented here illustrates the fact that "a reasons. First, it is perfectly adapted to object-
simplified model of a wind instrument produces oriented programming, of which one of the most
sound exceedingly similar to that of the actual sys- important characteristics is flexibility. Second, mod-
tem" (Keefe 1992). Contrary to many approaches ularity can be easily applied to the modeling of wind
where musical application of sound synthesis is the instruments owing to the existence of anticipative
principal objective, the objectives of physical mod- elements that connect to lumped elements. (These
eling are first to "test the accuracy of the implicit are both described later.) Finally, modularity holds
assumptions and structure of the underlying physi-much potential for evolution and a progressive ex-
cal model" with "simulation[s] of... known tension to the entire wind instrument family.
instrument-player system[s]," as Keefe (1992) re-
marks. The second objective is to allow the instru- A Representation Adapted
ment maker to have an idea of the sound produced to Object-Oriented Programming
by an instrument in a performance situation before
the construction of a prototype. The final objective In an object-oriented programming environment,
is to be able to design physically unrealizable vir- each object or class has associated attributes char-
tual instruments-for example, a transverse saxo- acterizing its structure as well as methods defining
phone (Bouasse 1929)-while remaining in a its behavior. This object communicates with others
particular timbre space and keeping the aspect of by its parameters and public properties, namely,
membership in a family of instruments. The pri- properties seen by the other objects, which must be
mary emphasis is placed on roughly modeling the defined according to a communications standard.
components of the instrument as well as signifi- On the other hand, the internal working procedure
cant performance controls. of the object, characterized by its parameters and
private
This article first presents a model of a single-reed properties, can be modified without disturb-
instrument based on the principle of modularity: the ing the entire model. This allows great flexibility,
instrument is built with interconnected elements. because one element of an instrument can be
This is made possible by an adapted representation changed while maintaining the integrity of the
of tubes. Next, the model of the single-reed mouth- whole model.
piece is detailed including the action of the player, For modeling single-reed wind instruments, four
including breath, tongue, and lips. The article thenobjects are distinguished: the mouthpiece and reed,
the tube, the tone hole with possible closing by a
finger or a key, and the bell (see Figure 1). Each
Computer element has one or two ports,
Music each one being char-
Journa
? 2003 Massachusetts In
acterized by an input pi, and an output p,,u. Accord-
Ducasse 59
This content downloaded from 154.59.124.94 on Fri, 14 Sep 2018 14:13:59 UTC
All use subject to https://about.jstor.org/terms
Figure 1. Four basic mod-
ules for building a single-
reed wind instrument.
Single-reed mouthpiece i
............cross
....... Tube with a varying
section
Tone-hole
Belwith )ell
pPcover "lPo --
-- ---- --
Decoupling of at
ing to the plane wave approximation Lumped
any Elements
port, Via
the average pressure p on the
thecross
Anticipative
section
Property S
of and
Tubes
the volume velocity u entering through this section
are given by A causal discrete-time system is called "anticipa-
tive" if the outputs can be calculated at least one
time step in advance, i.e., if the nth output sample
p(t) = Pin(t) + Po,,(t)
S S (1) can be calculated from the various variables at
sample n-1 and previous samples. Among the ele-
[u(t) = p- [pin(t)- pou,(t)l
ments constituting the wind instrument, the tube
where po is thepresents densitythis anticipative property
ofprovided
the its ai
sound. length is sufficient, that is, larger than c x T,, where
Each connection of two or more elements is real- T, is the sampling period chosen for the entire
ized by writing the continuity of pressure and con-simulation.
servation of volume velocity. Thus, this constitutes A system such as that shown in Figure 2 exhibits
the communication standard between the various anticipation on the left-hand side if the following
elements. can be written:
This content downloaded from 154.59.124.94 on Fri, 14 Sep 2018 14:13:59 UTC
All use subject to https://about.jstor.org/terms
Figure 2. The two-port net- and h2, are the four
work representing a por- discrete-time impulse re-
tion of tube of variable sponses of this two-port
section. Here, h*1, h12, h21, network.
p P
S 1,in
p- h* h 2,out
2 12
h* h*
S11 22
?? .
i -
S+h
66
i p*Z
Po
,b**
21
90**
1,out
0 6*6.01 6 . a&,D* *-f 50go-,0 0
P2.
2,in
9-06 0 *g 6.#e... . o 0 e..#...... 0
P,out(n) = i=0
h*1(j)p,,(n7=1
- j) + h*( j) pin(n - j)
In contrast to the tubes, the other causal ele
k22 k12 (3) ments are regarded as lumped because there
P2,ou(n) Si=o = j=1 h*(j )py,,(n anticipation.
- j) + h( j) are
They p~,(n - j) by a syste
characterized
This system does algebraic notor directly
differential equations. Consequent
check
shown in Equation lumped2. elements
But are because
directly connectedit one
other, they merge to form a single lumped ele
to the
to suppose assumption that Ih;l(0)This
of is why
slow I<<1
any and
connection
variation I h(0
can include only
of
as outlined in Ducasse 2002), the fol lumped element (Smith 1996). This rule imp
of variables can be made on the left-hand side of the existence of only two types of connection
Figure 2: First, it implies the junction of k tubes, which
linearly redistributes k output signals on k in
l,in = [1 + h;, (0) ] p owing to the continuity of pressure and conse
(4)
tion of volume velocity. Second, this rule im
Pou -= -- h1 (0) p,,n + Pou the connection of k tubes to a lumped element
Ducasse 61
This content downloaded from 154.59.124.94 on Fri, 14 Sep 2018 14:13:59 UTC
All use subject to https://about.jstor.org/terms
system of equations representing the inner behav- which is characterized by the radiation impedance
ior of this element includes its exchanges with of a cylinder of the same diameter (Levine and
these tubes. Schwinger 1948). From this radiation impedance,
This approach makes it possible to carry out a two linear causal invariant filters are calculated.
time-domain simulation according to a calculation Their common input is the output signal of the
performed over two cycles. At each time n, the out- tube at the open end. One filter calculates the re-
puts of all the pieces of tube are first calculated us-flected signal returning into the tube, and the other
ing their anticipative property. Then, the outputs delivers the transmitted radiated signal, assuming a
of the junctions and lumped elements, some pa- unipolar radiation.
rameters of which are controlled by the player in
the course of time, are calculated. Thus, each out-
put calculated is either the input of a tube or a ra- The Mouthpiece with Single Reed
diated signal. The numerical decoupling of the and the Mouth of the Player
lumped elements makes it possible to limit the
complexity of the equations to be solved, reducing
calculation time and numerical errors. The reed is modeled as a harmonic oscillator, usu-
ally used with fixed parameters (see, for example,
Wilson and Beavers 1974; Gazengel, Gilbert, and
Amir 1995). Here, these parameters are variables
Potential of Evolution and Generalization
that take into account the action of the lower lip as
to Other Wind Instruments
well as the shape of the slit. The reed interacts
with the tongue of the player. The flow from the
Modularity and object-oriented programming en- mouth to the inside of the mouthpiece depends pri-
able the evolution of these physical models and marily on the pressure difference between the
their generalization to other wind instruments by mouth and the mouthpiece and of the reed slit
allowing us to modify modules without changingopening. In the modeling, one of the difficulties is
the total structure or other elements. In addition to
to separate the quasi-static phenomenon-the
the modeling of other types of mouthpiece (e.g., flow-from the acoustic phenomenon. The effect of
Verge et al. 1994; Verge 1995 for the flute; Barjau the player's vocal tract is neglected (Backus 1985).
and Agull6 1989 for the double reed; Msallam et al. After the description of the model used, simulation
2000; van Walstijn and Campbell 1999 for brass in- results are given in numerical, graphical, and audi-
struments), one can imagine a non-linear modeling tory forms.
of portions of the tube of brass instruments, taking
into account possible shock waves formation
(Hirschberg et al. 1996; Msallam et al. 2000), effects
The Reed Coupled with Lower Lip System
due to a curved bore (Nederveen 1998a), and mod-
eling of the bell with higher modes calculation and
The reed is modeled as a damped spring-mass sys-
the directivity of the radiated field (Pagneux, Amir,
tem of variable mass M(4), damping G(4), and
and Kergomard 1996; Amir, Pagneux, and Kergo-
mard 1997; Kemp, Amir, and Campbell 2000). stiffness K(?, ,,e), where 4 is the reed slit opening
The detail of the modeling of the two lumped (see Figure 3), which results in the following differ-
ential equation:
elements in which the controls of the player are
important-namely, the mouthpiece with single
fx = ~ d2 dMd12
reed and the tone hole coupled with the finger/key
is provided hereafter and is illustrated by sound ex-
;.-=dt2 d dt- f (7)
amples. For these examples, a simplified model of +G C(5)d+ +K(eq)(-eq)
the bell has been chosen. Here, the bell is modeled
as a tube of varying cross section, the open end of Here, Req represents the reed opening at rest.
This content downloaded from 154.59.124.94 on Fri, 14 Sep 2018 14:13:59 UTC
All use subject to https://about.jstor.org/terms
Figure 3. Section of the
single-reed mouthpiece
and definition of the vari-
ous physical quantities.
0 20 40 60 80 (,a0 )
Pu
Pm m.
.... .... ..... ... ...
..
. . . . ...
.
.
..
La Reed,
flexural modes.
In fact, this equation includes On the
the other hand, a modal
influence of analy-
both the lower lip of the sis player
would show thatand the phenomenon
the shape of rolling
ofup
is not
the slit. In particular, 4eq inevitably present
depends on andthe
that itpressure
would thus be
necessary
exerted by the player's lip. Theto consider
damping a much more iscomplex
consid- rela-
ered to be mainly caused byof the
tionship contact.lip, because
This point when
would require much
players grip the reed with moretheir
thorough work.
teeth, they inevi-
tably obtain a "squeak" where the reed seems to
oscillate on one of its eigenmodes (Wilson and
Beavers 1974; Boutillon The
and Gibiat
Tongue 1996).
and Its Interaction On
with the the
Reed
other hand, the mass M and the stiffness K are pri-
marily related to the density
The tongue of
plays the reed
an important role and itsIt lon
in phrasing.
gitudinal Young's modulus. (Consider, for example
is modeled by a damped spring-mass system, as
the Fibracell reed, which is made
shown in Figure 4,of an
whose orthotropic
equilibrium position is
composite material; see controlled.
Shaffer 1982.)
It can touch the reed These pa-
with a perfectly
rameters also depend on soft
the geometry of both the
shock: just after the shock, the reed and the
reed and the mouthpiece. The variability of
tongue form a single system whose momentumthe pa-
is
rameters according to 4 the
is sum
in particular
of the two momenta beforeused
the shock.to
An take
into account the "rolling up"
adhesion modelphenomenon
is introduced for the separation so
whereby the reed progressively closes
that the tongue and the reed on the
separate only if the
mouthpiece. force that they mutually exert on each other is
In Equation 7, the sum ofhigherthethan a external forces
given threshold. This threshold, Ias f,,
includes the force due to the
well as the difference
other parameters of the in pressure
tongue model,
between the two sides of the
can vary reed
in time in order(an equivalent
to obtain various effects
surface S, (4) is considered) and also the action of
of phrasing.
the tongue. Even if the results obtained seem prom-
ising from an auditory point of view, many ques-
tions remain on the mechanical behavior of the
Air and Its Interaction with the Reed
"reed + lay + lip" system (see, for example, Stewart
and Strong 1980; Gazengel et al. 1995). A modal
Conservation of Flow
analysis using finite elements on a viscoelastic
model would tend to prove on the one hand that The pressure difference between the mouth of the
the torsion modes play a role as important as the player pm and the mouthpiece volume p, acts on
Ducasse 63
This content downloaded from 154.59.124.94 on Fri, 14 Sep 2018 14:13:59 UTC
All use subject to https://about.jstor.org/terms
Figure 4. Simplified rep-
resentation of the
reed + lower lip system
and the tongue.
Keq
Results
This content downloaded from 154.59.124.94 on Fri, 14 Sep 2018 14:13:59 UTC
All use subject to https://about.jstor.org/terms
Figure 5. Time-frequency
analysis (spectrogram) of a
"squeak" on a virtual in-
strument (Sound Example 1).
7
Design of Imaginary In
-mo
and the "Radiator" Net
On a musical level, on
- ? P . - ..
tions of .. - ..
physical .
modeli
of physically unrealiza
2 - is the transverse saxoph
study shows that one m
"Helmholtz signal" who
the ratio of the lengths
and Kergomard 1995). D
o..! .. .... ........I of the rectangular signa
and (-NPo) during TI/(N
ize such an instrument
piece volume equal to
o a : -l :2:::1. :':- 2 L'IJ
represented i"-
by Sound E
ment with the theory, a
close to 5, the sound obt
01 0.r 1 1. t(s) saxophone. One can also
mouthpiece to a "centr
tube network (Sound E
namic plays a prominent part, as is well know
players and instrument Importance
makers, and of its the Tong
drivin
delicate on the virtualTheinstrument
player can as obtain
in realit a
Various modes of oscillation,
articulations more by or less st
using th
are obtained when theing damping of the of the reed
tongue gives co
with the lower lip is varied,
sounds as illustrated
produced. It canby
Example 2 and Figure model6.is sufficient to obtain sound effects charac-
With identical parameters, different
teristic of single-reed instruments (Sound modes
Examples c
be obtained depending on
1, 6, 7, andwhat
8 for a given occurred
resonator, and Tracks previ
9 et
ously. In Sound Example
seq. for2, the
complete damping
models with tone holes). G is g
ually increased by a factor of 200 while all oth
parameters are held constant. Three success
modes of oscillation The Tone Hole andbut
appear, the Finger of the Player
when the d
is decreased symmetrically, the oscillation r
stable on the lower mode,
The Model which was obtain
tially with a high damping. This shows, as con
Many works
firmed by players, that theonattackopen and closed
hastone an
holes have
imp
influence on the sound quality
been published obtained.
(e.g., Nederveen, Va
Jansen, and Van
Ducasse 65
This content downloaded from 154.59.124.94 on Fri, 14 Sep 2018 14:13:59 UTC
All use subject to https://about.jstor.org/terms
Figure 6. Time-frequency 2 for the three modes of ure, adjustments are rigor-
analysis (spectrogram) and oscillation (partials 9, 7, ously identical; only their
phase diagrams (pm -Pc) and 5) and the transient past differs.
flow uc of the sound re- between the second and
corded on Sound Example third modes. On each fig-
.. ..... .. . .. . . . .
. .................
V4.~
S-hi
.. .2 - ": : . . . .. ? . . ?
mm ..... ....... ... .........
m r ...........
This content downloaded from 154.59.124.94 on Fri, 14 Sep 2018 14:13:59 UTC
All use subject to https://about.jstor.org/terms
Figure 7. Definition of the average perimeter), w is its hole, ua outgoing by the external acoustic pressure
various physical quantities "width" (the distance opening, and up due to the is considered negligible.
concerned in the closing cover / body of the instru- displacement of the cover. The radiated output signal
system (key or finger) of a ment), and d is its The acoustical pressure at is taken proportional to
tone hole: 1 is the "length" "depth." The flows are u, the input of the tone hole the flow ua, assuming a
of the opening (in fact, the entering into the tone is denoted by p, and the unipolar radiation.
Results
p
Ducasse 67
This content downloaded from 154.59.124.94 on Fri, 14 Sep 2018 14:13:59 UTC
All use subject to https://about.jstor.org/terms
Figure 8. Time-frequency end of the sound, the pro-
analysis (spectrogram) of gressive opening of the key
Sound Example 9 obtained makes it possible to ob-
from a cylindrical tube tain a glissando.
with a tone hole. At the
f (Hz)
This content downloaded from 154.59.124.94 on Fri, 14 Sep 2018 14:13:59 UTC
All use subject to https://about.jstor.org/terms
Barjau, A., and J. Agull6. 1989. "Calculation of the Start- Hirschberg, A., et al. 1996. "Shock Waves in Trombone."
ing Transients of a Double-reed Conical Woodwind." Journal of the Acoustical Society of America 99(3):
Acustica 69:204-210. 1754-1758.
Bouasse, H. 1929. Instruments a vent, tome I. Paris : Bib-
Keefe, D. H. 1983. "Theory of the Single Woodwind To
liotheque Scientifique de L'Ingenieur; Reprint, Paris:
Hole" and "Experiments on the Single Woodwind
Blanchard, 1986. Tone Hole." Journal of the Acoustical Society of
Boutillon, X., and V. Gibiat. 1996. "Evaluation of the
America 72(3):676-699.
Acoustical Stiffness of Saxophone Reeds underKeefe,
Playing
D. H. 1992. "Physical Modeling of Wind Instru-
Conditions by Using the Reactive Power Approach."
ments." Computer Music Journal 16(4):57-73.
Journal of the Acoustical Society of America 100(2):
1178-1189.
Kemp, J., N. Amir, and D. M. Campbell. 2000. "Calcula
tion of Input Impedance Including Higher Modes."
Cadoz, C., A. Luciani, and J.-L. Florens. 1993. "CORDIS-
Proceedings of the Fifth French Congress on Acoustic
ANIMA : A Modeling and Simulation System for
Sound and Image Synthesis-The General Formalism." Lausanne: E.P.F.L., pp. 314-317.
Computer Music Journal 17(1):19-29. Kergomard, J. 1995. "Elementary Considerations on
Castellengo, M. 1999. "Acoustical Analysis of Initial Reed-Instrument Oscillations" In A. Hirschberg, J. Ker
Transients in Flute Like Instruments." Acustica 85: gomard, and G. Weinreich, eds. Mechanics of Musical
387-400. Instruments. Vienna: Springer-Verlag, pp. 229-290.
Cook, P. 1992. "A Meta-Instrument Physical Model, and H., and J. Schwinger. 1948. "On the Radiation of
Levine,
a Meta-Controller for Real-Time Performance Con- Sound from an Unflanged Circular Pipe." Physical Re-
trol." Proceedings of the 1992 International Computerview 73:383-406.
Music Conference. San Francisco: International Com- Msallam, R., et al. 2000. "Physical Model of the Trom-
puter Music Association, pp. 273-276. bone Including Nonlinear Effects. Application to the
Cullen, J. S., J. Gilbert, and D. M. Campbell. 2000. "BrassSound Synthesis of Loud Tones." Acustica 86(4):725
Instruments: Linear Stability Analysis and Experi- 736.
ments with an Artificial Mouth." Acustica 86:704-
Nederveen, C. J. 1998a. "Influence of a Toroidal Bend on
724.
Wind Instrument Tuning." Journal of the Acoustical
Dalmont, J.-P., and J. Kergomard. 1995. "Elementary
Society of America 104(3, Part 1):1616-1626.
Model and Experiments for the Helmholtz Motion of
Single Reed Wind Instrument." Proceedings of the Nederveen, C. J. 1998b. Acoustical Aspects of Woodwind
1995 International Symposium on Musical Acoustics. Instruments. Rev. ed. DeKalb, Illinois: Northern Illi-
Dourdan, France: Soci6t6 Frangais d'Acoustique, pp. nois University Press.
114-120. Nederveen, C. J., J. K. M. Jansen, and R. R. Van Hassel.
1998.
Dubos, V., et al. 1999. "Theory of Sound Propagation in a "Corrections for Woodwind Tone-Hole Calcula-
tions."
Duct with a Branched Tube Using Modal Decomposi- Acustica-Acta Acustica 84:957-966.
tion." Acustica 85:153-169. Pagneux, V., N. Amir, and J. Kergomard. 1996. "A St
Ducasse, E. 2002. "An Alternative to the Traveling-Wave
of Wave Propagation in Varying Cross-Section Wav
Approach for Use in Two-Port Descriptions of Acous-
guides by Modal Decomposition. Part. 2: Theory
tic Bores." Journal of the Acoustical Society of Amer-
Validation." Journal of the Acoustical Society of
ica 112(6):3031-3041. America 100(4, Pt. 1):2034-2048.
Gazengel, B., J. Gilbert, and N. Amir. 1995. "Time Do-
Shaffer, D. W. 1982. "Reed Construction." U.S. Pat
main Simulation of Single-Reed Wind Instrument.
No. 4,355,560.
From the Measured Input Impedance to the Synthesis
Smith, J. 0. 1986. "Efficient Simulation of the Reed-
Signal. Where are the Traps?" Acta Acustica 3:445-
472. and Bow-String Mechanisms." Proceedings of the 1
Grand, N., J. Gilbert, and F. Lalo.. 1997. "Oscillation International Computer Music Conference. San Fran
Threshold of Woodwind Instruments" Acustica 83(1): cisco: International Computer Music Association,
137-151. 275-280.
Ducasse 69
This content downloaded from 154.59.124.94 on Fri, 14 Sep 2018 14:13:59 UTC
All use subject to https://about.jstor.org/terms
Stewart, S. E., and W. J. Strong. 1980. "Functional Model Verge, M.-P., et al. 1994. "Jet Formation and Jet Velocity
of a Simplified Clarinet." Journal of the Acoustical So- in a Flue Organ Pipe." Journal of the Acoustical Soci-
ciety of America 68(1):109-120. ety of America 95(2):1119-1132.
Van den Berg, J., J. T. Zantema, and P. Doornenbal, Jr. Verge, M.-P. 1995. Aeroacoustics of Confined Jets, with
1957. "On the Air Resistance and the Bernoulli Effect Applications to the Physical Modeling of Recorder-
in the Human Larynx." Journal of the Acoustical Soci- Like Instruments. Eindhoven: Technische Universiteit
ety of America 29:626-631. Eindhoven.
van Walstijn, M., and M. Campbell. 1999. "The Wave Wilson, T. A., and G. B. Beavers. 1974. "Operating
Digital Filter Brass Mouthpiece Model." Acustica Modes of the Clarinet." Journal of the Acoustical Soci-
85:S77. ety of America 56(2):653-658.
This content downloaded from 154.59.124.94 on Fri, 14 Sep 2018 14:13:59 UTC
All use subject to https://about.jstor.org/terms