Download as pdf or txt
Download as pdf or txt
You are on page 1of 4

Audio Engineering Society

Convention Paper 8997


Presented at the 135th Convention
2013 October 1720
New York, NY, USA
This Convention paper was selected based on a submitted abstract and 750-word prcis that have been peer reviewed by at least
two qualified anonymous reviewers. The complete manuscript was not peer reviewed. This convention paper has been
reproduced from the author's advance manuscript without editing, corrections, or consideration by the Review Board. The AES
takes no responsibility for the contents. Additional papers may be obtained by sending request and remittance to Audio
Engineering Society, 60 East 42nd Street, New York, New York 10165-2520, USA; also see www.aes.org. All rights reserved.
Reproduction of this paper, or any portion thereof, is not permitted without direct permission from the Journal of the Audio
Engineering Society.

An Objective Comparison of Stereo


Recording Techniques Through the Use of
Subjective Listener Preference Ratings
Lim, Wei
University of Michigan, Ann Arbor, Michigan, 48109, USA
limwei@umich.edu
ABSTRACT
Stereo microphone techniques offer audio engineers the ability to capture a soundscape that approximates how one
might hear realistically. To illustrate the differences between six common stereo microphone techniques, namely
XY, Blumlein, ORTF, NOS, AB and Faulkner, I asked 12 study participants to rate recordings of a Yamaha
Disklavier piano. I examined the inter-rating correlation between subjects to find a preferential trend towards nearcoincidental techniques. Further evaluation showed that there was a preference for clarity over spatial content in a
recording. Subjects did not find that wider microphone placements provided for more spacious-sounding recordings.
Using this information, this paper also discusses the need to re-evaluate how microphone techniques are typically
categorized by distance between microphones.

1.

INTRODUCTION

1.1.

Background

Stereo microphone and playback techniques were


specifically developed to allow for proper representation
of a normal listening field one perceives. There are
many commonly employed techniques that evolved
based on the principles of human auditory perception.
Recent papers published in major journals have focused
on the advancement of encoding and decoding in
playback systems. Detailed, further research in twomicrophone stereo recording techniques, beyond the

initial conceptualization of the techniques, is lacking.


Olive coined the term: circle of confusion where by
any one element of the entire sound reproduction
procedure, namely recording, mixing, duplicating and
playback, could be a cause for misrepresentation in
reproduced sounds [1]. It is therefore dire that we
understand every step of the process, including the very
first stage of sound reproduction the recording.
1.2.

Aim and Hypothesis

A survey of existing literature shows that there is a


shortage of direct comparisons in the literature with
regard to how stereo microphone techniques differ from
each other, and the advantages of using a technique over

Lim

Comparison of Stereo Microphone Techniques

another. This paper hopes to fill the gap by examining


the results of a formal, blinded listening test, where
participants rate their relative preferences of extremely
identical recordings using different stereo microphone
techniques, done in a highly-controlled environment.
The microphone techniques examined in this
experiment were XY, Blumlein, ORTF, NOS, modified
Faulkner (at 60cm apart) and AB (at 60cm apart), with
recordings performed using a pair of AKG C414-ULS.
These techniques attempt to simulate human auditory
perception in hopes of being able to accurately recreate
a combination of sounds that forms an immersive
environment, also known to us as a soundscape [2]. Of
these microphone techniques, we can classify them into
three categories coincidental (XY and Blumlein),
near-coincidental (ORTF and NOS) and spaced
configurations (modified Faulkner and AB).
In this study, I made three hypotheses. Firstly, I
predicted that there would be a preference shown
towards the near-coincidental techniques because nearcoincidental techniques blend the strengths of both the
coincidental and spaced techniques. The second
hypothesis was that the wider microphone placements
(distance between capsules) do not always yield a
higher rating in terms of perceived spaciousness. Lastly,
I hypothesized that there is a trade-off in semanticpreferential ratings between clarity and spaciousness,
with listeners generally preferring clarity captured over
spatial content in their recordings.
2.

METHODS

2.1.

Participants

Subjects have between two and nineteen years of


experience in critical listening. Previous studies by
Toole [3] have shown that trained listeners are more
discerning, but not less representative of the larger
audience. All participants were not compensated for
their time, and additionally, were told that their
participation was completely optional and on a
voluntary basis.
2.2.

Materials

In order to accurately determine the preferences


amongst questions per recording technique, a blinded
experiment set up was created in the Cycling 74s
Max/MSP environment. The Max/MSP patch allowed
users to submit their relative preferential ratings of each

of the samples played. Six questions were asked about


each of the sample recordings that were presented two
at a time. The questions were phrased such that they
elicited responses specific to the following six
attributes: representation of the instrument recorded,
wideness of the space perceived, quality of localization,
clarity of notes (in terms of attack and articulation),
depth of space perceived and a general preference. For
each question, subjects used an 11-point semantic rating
scale to show their preferences (0 = not at all, 10 =
extremely).
The presented music was performed on a Yamaha
Disklavier MX100A, recorded in the Rolston Concert
Hall at The Banff Centre in Alberta, Canada. The
upright piano was sent MIDI information triggered
through a digital audio workstation. The mechanical
playback ensured that there was a consistent and highly
identical performance for all of the recordings I
performed. As such, it allowed me to keep the
microphone, the microphone pre-amplifier, the cables,
the stands, and the position from the piano exactly the
same, while only changing the stereo techniques set up,
thus minimizing confounding factors in this study.
2.3.

Procedure

During the study, participants were first briefed on the


purpose of this experiment. They then were introduced
to the experimental interface, and proceeded with the
test. The study required them to answer a series of 6
questions, repeated for each of the 3 categorical pairs of
microphone techniques. A sequence with a total of 18
questions was set up such that no two participants will
receive the same order of questions and choice
presentation. I had randomized the test presentation
order to avoid possible biases and threat to internal
validity of the experiment due to order effects.
After the first set of questions was completed, another
set of questions was generated based on the participants
submitted results. These questions would only contain
sound samples that were previously rated the highest for
their respective technique categories (coincidental, nearcoincidental
and
spaced)
per
attribute
(representativeness, wideness of space perceived,
localization, definition in attack, depth of space
perceived, and general preference).
Each subject then answered some demographical
questions. These questions included the number of years
they have had formal training in audio or worked

AES 135th Convention, New York, NY, USA, 2013 October 1720
Page 2 of 4

Lim

Comparison of Stereo Microphone Techniques

professionally in the audio industry, their age, the


gender they identified themselves with, the musical
genre that they work in or listen to most, and the
listening environment that they were in while
participating in this experiment. These questions were
designed to find out if moderating variables were at
play.
3.

RESULTS

To test the first hypothesis, the general preference rating


of each microphone technique within each of the
categories (coincidental, near-coincidental or spaced)
was added together. They were then analyzed for the
differences between group means.
I also measured the correlation between the sense of
spaciousness perceived in a recording, and the category
it belongs to. Spaciousness was measured by adding the
ratings for the wideness and depth of space perceived in
each recording sample.
The clarity measure was a sum of attack and ability to
localize. All statistical results were tested for normality
and had a Greenhouse-Geisser correction where
relevant.
An ANOVA test shows that across the three categories
of microphone techniques, there is a significant
difference between General Preference ratings of NearCoincidental techniques versus Coincidental and
Spaced, with p=.02 and p=.03 and mean difference of
1.41 and 1.31 respectively.
The second hypothesis is upheld, given that there is no
significance (p>.85) between the spaciousness
perceived in a recording, and how far apart the
microphone placements were between capsules.
Listeners also generally preferred Clarity over
Spaciousness, with results showing a correlation
between the former category ratings (Pearsons 2-tailed
with p=.01) and a lack of correlation between the latter
(p=.06), and General Preference ratings. There is,
however, no evidence of a trade-off between Clarity and
Spaciousness.
4.

DISCUSSION

Statistical results have shown that listeners, indeed,


prefer near-coincidental techniques, as predicted in the

first hypothesis. According to Dooley [4], such


techniques provide for good localization. This thus
serves to solidify the third hypothesis, being that
listeners do lean toward clarity in a recording.
With the second null hypothesis rejected, it is worth reexamining the way in which we think and learn about
microphone techniques. Stereo microphone techniques
have traditionally been categorized by the distance
between the microphone capsules, as defined by the
techniques placements. The results suggest that other
factors, such as microphone polar patterns, could be at
play. In the Blumlein technique, which is classified as a
coincident pair technique, specifies figure-8 polar
pattern the microphones. This probably was the reason
why Blumlein recordings were found to be highly
similar to both of the Spaced pair recordings (p<0.01).
5.

LIMITATIONS AND FUTURE DIRECTION

The materials in this experiment have been arbitrarily


selected. I recognize that the microphone choice may be
said to be less ideal for recording classical piano music,
not to mention the upright piano. Yet, the versatility and
availability of the microphone was taken into
consideration, and thought to be most fitting in this
instance. In spite of the arbitrary selection of materials,
one should note that they were kept constant and thus
there is high internal validity within this experiment.
The only variable was the microphone technique. I
therefore do not expect any confounding factors in the
experiment conducted.
I am also aware of the small sample size I have attained
in this experiment, thus leading to a slightly reduced
generalizability. In the mean time, however, the found
results still speak given the high internal validity and
appropriate use of statistical tests.
Future research may be built upon the interface created
for testing subjects. The interface allows for random
presentation order, blind tests and quick access to
results. As such, it could be beneficial to expand beyond
the six stereo microphone recording techniques used in
this experiment. It would also be interesting to compare
binaural recordings against traditional stereo recordings.
Data should also be collected with a different
performance hall, different selection of music and
different microphones.
This experiment has shown that there are indeed valid
and objective ways of testing for preferences between

AES 135th Convention, New York, NY, USA, 2013 October 1720
Page 3 of 4

Lim

Comparison of Stereo Microphone Techniques

recording techniques. The systematic constants in the


test set up and experiment interface that allowed for
blind testing, are strengths of this experiment that I
think should be displayed in other formal experiments
involving recording techniques and auditory perception.
6.

ACKNOWLEDGEMENTS

I would like to thank Jason Corey for mentoring me


through this entire project, and the wonderful people at
The Banff Centre, especially Peter Cook, who assisted
me during the formulation of the idea and for allowing
me to have full access to their resources to record the
necessary source material.
The generous help of all who have participated in my
study and provided feedback towards the current
interface is also greatly appreciated. The collection of
materials would not have been possible without the help
of Benjamin Gendron-Smith, Denis Martin and
Winfried Lachenmayr.
7.

REFERENCES

[1] Olive, S. (1990). The preservation of timbre:


Microphones. Loudspeakers, sound sources and
acoustical spaces. In Proceedings of the AES 8th
International Conference, May 3-6. Audio
Engineering Society.
[2] Rumsey, F. (2002). Spatial quality evaluation for
reproduced sound: Terminology, meaning, and a
scene-based paradigm. Journal of the Audio
Engineering Society 50(9). 651 666.
[3] Toole, F. E. (1981). Listening tests turning
opinion into fact. Presented at 68th Audio
Engineering Society Convention, Los Angeles.
Preprint 1766.
[4] Streicher, R., and Dooley, W. (1985). Basic Stereo
Microphone Perspectives-A Review. Journal of the
Audio Engineering Society 33, 7/8, 548556.

AES 135th Convention, New York, NY, USA, 2013 October 1720
Page 4 of 4

You might also like