SOUND FIELD SIMULATION
auralization, an analogy to vizualization. Through
ABSTRACT
auralization, it is possible to identify the objective
parameters that correspond to certain subjective
This document explains the basics behind acoustic
reactions experienced by listeners.
computer simulation. It includes details on how human
hearing uses several techniques to localize sound
ACOUSTIC SIMULATION
sources, how we can simulate factors that influence
human auditory perception with computer software,
To determine the location of a sound source, our brain
and how we can reproduce the listening experience for
uses a number of different cues, among the following
a space that has not been built.
[Malham 1998]:
It also includes a case study that analyzes numerical
a. Sound reaches an individual s two ears at
parameters and creates a sound simulation of a space
different times. As long as the source of the sound is
that allows the listener to sujectively  grade the
not directly behind or in front of a listener, the sound
acoustical qualities.
will arrive at one ear before it arrives at the other. The
time difference is known as  Interaural Time Delay
INTRODUCTION
(ITD). This effect occurs only at frequencies where the
wavelength is less than twice the distance between the
Over the last several decades, the process of predicting
ears; by this technique alone, humans are unable to
the acoustics of a room in advance of its construction
determine the location of sound with longer
has advance from an art into a controlled and exact
wavelengths.
process. In the last few years computer models have
grown from being just a supplemental tool to becoming
b. Sound reaches an individual s two ears at
a full substitute of earlier techniques and a superior
different levels. When a sound source is located to the
design method.
side of a receiver, the sound reaches one ear directly.
The other ear receives sound only after it has diffracted
Devices capable of realistically simulating typical room
around the head. Sound arriving at the occluded ear
sound fields will prove to be a useful tool for
will therefore be quieter.
understanding the acoustical qualities of spaces. They
will help the acoustician (as well as people outside of
c. The human brain can distinguish sound
the field) make decisions on projects that address
position relative to a phenomenon called the Head
acoustical issues.
Related Transfer Function (HRTF). HRTF is a
frequency dependent response that varies with source
Results can be visualized and analyzed much better
location, and is based on the shape of the head and the
than before because the computer model contains more
external part of the ears.  When the source gives an
information than a set of measurements done in a scale
ambiguous ITD, this is the brain s main position-
model or than a single number rendered by formulae.
sensing mechanism [Malham 1998].
Today, computer models have become reliable and
efficient design tools for the acoustic consultant, and
d. By moving one s head, a listener can vary the
 the results of a simulation can be presented not only
ITD and adjust the HRTF between the ears, giving the
for the eyes but also for the ears with techniques for
brain more information to determine the sound source
auralization [Rindell 2000] .
location.
By modeling the significant acoustical parameters of a
The best system to recreate the acoustic qualities of a
design, we can preview a proposed acoustical solution.
space is a surround sound alternative known as
The technique of using computer software to predict
Ambisonics. It offers features impossible to realize
and recreate sound in a given room has been called
through other methods. With this system it is possible
to capture a sound event (such as a musical yet built. This software allows the acoustician to define
performance) and replay it such that, as far as possible, plane surfaces in order to build a three-dimensional
the original sound and acoustical environment of the model of a space in a computer. The acoustician then
original performance is faithfully recreated. assigns to each surface certain material characteristics:
Ambisonics satisfies simultaneously as many as the degree to which it absorbs, reflects and deflects
possible of the mechanisms used by the brain/ear to sound waves. A source can then be defined: where it is
localize sound. located, how its directivity is shaped, and how powerful
the source signal is.
An additional practical benefit is that the realistic
listening area for Ambisonic Surround Sound is larger The code computes the path of sound from a source to a
than that of conventional stereo [Elen 2001]. receiver. As the sound travels from the source to the
receiver, it may reflect off of walls, defract around
Ambisonics equipment generates a 4-channel signal,
edges, or arrive at a receiver position directly.  These
known as the B-Format, that contains all the
paths are utilized to simulate and predict the acoustic
information in the soundfield (direction, delay, sound
qualities of the space [Markham 2002].
intensity, etc). These four channels record the event
into left-right, front-back and up-down information CATT-Acoustic software relies on the real geometric
plus a mono reference signal [Elen 2001]. Figure 1 layout of the space and specific properties (sound
shows the B-Format signal. absorption and diffusion) of materials within that space.
The software uses a technique similar to ray tracing
known as  randomized tail-corrected cone tracing in
order to perform acoustics calculations and simulations
[Dalenback 2002]. Simulations are run with one source
sending its signal through the room. Each point receiver
shows the values for reverberation time (the time in
seconds that it takes for a sound to reduce in sound
pressure level by 60dB after the sound source has been
silenced) and other variables at a sampled listening
post. CATT-Acoustic also illustrates the decay curve at
each receiver point, which indicates how the sound died
out.
Figure 1 B-Format signal
SIMULATION OF A BUILT SPACE
When replayed the ambisonic signal is processed and
Before aurally simulating the acoustical environment of
fed from a decoder to each speaker independently to
an unbuilt space it is important to affirm the acuracy of
create a horizontal surround. (A minimum of four
the software. To do this we created and processed a
speakers are required.) Each signal contains all the
model of an existing space to compare the numerical
elements of the original recording but in different
parameters against actual measurments taken at the
ratios, working together to recreate the ambience and
physical space.
acoustics of the original space.
The Ruth Shapiro Theater is located at Brandeis
The important thing to note is that there is no need to
Univeristy in Waltham Massachusetts. This 250-seat
consider the actual details when doing B-Format
hall will accommodate productions by the
recording, synthesis or reproduction: if B-Format
Undergraduate Theater Collective and lectures, among
specifications are followed and suitable
other events. Figures 2 and 3 show an interior image of
loudspeaker/decoder setups are used,  all will be well
the space and a section of the computer model,
[Malham 1998].
respectively.
CATT-Acoustic [Dalenback 2002], an acoustical
prediction software, is capable of generating a B-
Format room response of a room which has not been
recommendations from CATT-Acoustic. The values
used derive from personal judgment from visual
analysis of the surfaces.
The computer predictions differ from the average
measurement results by the same magnitude as any two
individual field measurements. All are within the 5%
subjective difference limen range [Gibbs & Oldham]
except in the 1KHz frequency band. Thus, we can
deduce that the accuracy of the best computer
calculations is roughly as good as two consecutive
measurement results and can be considered acceptable.
See graph 1.
Reverberation Time Comparison Graph
Figure 2 Interior image
Real
1.3
Predicted
5% Limen
1.2
1.1
1.0
0.9
0.8
0.7
125 250 500 1K 2k 4K
Figure 3 Ruth Shapiro Theater computer model
Hertz
Graph 1 Reverberation comparison.
The computer model was initially set up for verifing the
results from the reverberation time calculations.
SIMULATION OF AN UNBUILT SPACE
Generally, larger rooms have longer reverberation
After validating the reverberation time computer
times simply because it takes longer for the sound to
prediction, the next step it to generate a simulation of
travel around the room. Rooms with shorter
the environment of a space which has not yet been
reverberation times tend to be better suited for speech,
built; in simple terms, an acoustic test drive.
while rooms with reverberation times around 2 to 3
The need for an  acoustic test drive arose in the
seconds are good for solo or orchestral music. For
development of a large atrim space. This enormous
organ music and chant, Rooms with longer
glass structure of 800,000 sqft. is planned as a multi-
reverberation times are preferred [Cowan 2000].
purpose space, where large dining events with music
Reverberation time (RT) is far from the only metric
can be held. The lack of absorptive materials in the
used to judge the acoustics of spaces. However, the RT
original concept, resulting in an unacceptably long
is a good starting point since it is a central parameter in
reverberation time, was a concern shared by the owner,
many applications of room acoustics. In order for a
architect and acoustic consultant from the early stages
room to achieve appropriate room acoustics conditions,
of the design procsess.
most acousticians would agree than a room must have
The single most important variable that influences the
an appropriate RT.
acoustics in the atrium is the number of sound
The material characteristics are needed for recreating
absorbing units, or sabins, in the space. A sabin can be
the acoustical environment through the computer. The
defined as a totally absorptive area of 1 sqft. The
absorption performance data of the materials were
absorption coefficient ranges from 0.01 to 1; fuzzy
taken from Cavanaugh [1999]  Acoustical control in
porous materials (velvet, glass fiber insulation) are on
Buildings . The data on diffusion were deduced from
the higher region of the range while hard and dense
physical reasoning following the scattering coefficient
materials such as marble have a low coefficient. This
Seconds
study provided the tools to establish objective goals for T-1: Inclusion of sound absorptive treatments on one of
the acoustical treatment in the proposed design. the walls and some structural elements, average
absorption coefficient of 0.18 and a reverberation time
CATT-Acoustic was used to calculate the room
of 3.2 seconds.
response and acoustic signature of three different
versions of the modeled space. Figure 4 shows the T-2: treatments as proposed in Scheme T-1, plus an
acoustical treatment in color orange on each version. additional area of sound absorption, around 4,300
sabins, to represent the amount of treatment that is
believed will achieve the goals of the Owner. This
version has an average absorption coefficient of 0.25
and a reverbaration time of 2.7 seconds. The actual
location of this extra material is on the wall located
directly in front of the previously treated wall.
Multivolver, a software application that works in
conjunction with CATT-Acoustic, processes the room
impulse response from the computer model, the source
material (eg. music or speech) and the loudspeaker
layout of the test room where the auralization will be
reproduced. An Ambisonic reproduction of the
complete sound field one would experience during an
T-0
actual event is generated and allowes reasonable
comparisons of different acoustical conditions,
replicating what a listener would hear in the projected
space for each of the given conditions.
As source material, recordings of the sound of groups
in a dining environment are used. The recording was
done in a small, rather non-reverberant restaurant
environment so that the room acoustics of the recording
space did not influence the final auralization. The
recordings were processed into the acoustic computer
model so that it simulates the environment of 500
diners, situated at tables around the main volume.
T-1
The particular auralization technique developed for this
project allows the listener to hear the various conditions
in the laboratory test room without having to wear
headphones. The sound surrounds the listener as the
sound would in the real space.
At the acoustic consultant´s laboratory, seven different
types of events in the three differently treated spaces
were developed. The laboratory is a semi-anechoic
room treated with absorptive ceiling and wall panels as
well as carpeted floor; the room characteristics prevent
any reflection from the real space to hamper the virtual
T-2 simulation.
For the simulation, four B-Format decoded signals
were fed to four loudspeakers situated in a horizontal
Figure 4 Treatment schemes
square array. This created a pantophonic (360°) system
T-0: all hard surfaces, no acoustical treatment, average that recreated the sound stage of the future atrium. This
absorption coefficient (total number of sabins evenly helped key persons participating in this project make
distributed over all surfaces) of 0.03 and a their own judgment of the relative effect of different
reverberation time of ~14 seconds (never actually amounts of sabins in the space. Table 1 is a table given
considered, but useful for comparison). to the participants to select any of the 21 auralization
available.
T-0 T-1 T-2
subjectively judge acoustics in space, participants must
Estimated Reverberation Time (sec) ~ 14.0 3.2 2.7 be able to listen to how the space will sound.
0.03 0.18 0.25
Estimated Average Absorption
The auralization techniques offer the possibility to use
the ears and listen to the acoustics of a room during the
Auralization Choices:
design process. Several acoustical problems can be
Banquet, 500 diners only
detected by the ears, whereas they may be difficult to
Banquet, 500 diners, plus classical
express with a parameter that can only be calculated.
Banquet, 500 diners, plus jazz
Using these tools the acoustician can communicate the
Banquet, 500 diners, plus rock
acoustic consequences of a design to the client/architect
Banquet, 500 diners, plus solo
effectively. This technique can be used very early in the
Banquet, 500 diners, amplified
project to achieve the desired results.
speech
Typical use, just passers-by
Nevertheless, it is important to note that the technique
described in this paper relies on a model. With
Table 1 Auralization table
listeners focusing on acoustical tests in a visually
The result was a fair and accurate representation of the
artificial environment, the experience can obviously
various acoustical conditions and ambience. A clear
never be totally representative. The acoustician must
difference was perceived by the listeners; some even
always remind the participants of the physical and
experimented by having conversations while in the
psycho-acoustic limitations of the model.
laboratory simulation such as they would have at a
cocktail party. With versions T-0 and T-1 it was
ACKNOWLEDGMENT
difficult to perceive the music playing in the
This work was supported by Acentech Incorporated.
background, and it was very uncomfortable to carry out
Ther authors wish to thank Carl Rosenberg and Leslie
a conversation. Although T-2 was by no means quiet
Norford for their guidance and support on this research.
(remember it is 500 diners plus music) it was much
more comfortable to speak to individuals seated on the
REFERENCES
opposite side of the table, and the music can be clearly
distinguished. It also has the added advantage the extra
ACENTECH and Cowan, James. Architectural
absorption has of noticeably reducing the intensity of
Acoustics Design Guide, McGraw Hill, 2000.
the ambient noise levels.
Markham, Benjamin.  Renovation of Sound. Thesis,
The feedback from the listeners was very positive and
Bachelor of Science in Engineering, Princeton
they commented favorably on the help this simulation
2002
rendered for the decision-making process. Version T-
Cavanaugh, William and. Wilkes, Joseph.
2 with the extra 4,300 sabins of treatment was selected
 Architectural Acoustics  Principles and
as the most desirable option and is now incorporated on
Practice , Wiley & sons, 1999.
the design.
Dalenbach, B.I.,  CATT-Acoustic User s Manual ,
CONCLUSION
2002.
Previously, acoustics could be measured, quantified
Elen, Richard,  Ambisonics: The Surround
and charted, but with the help of auralization the
Alternative 2001, [article on-line], available from
subjective character of sound can now be simulated and
http://www.ambisonic.net/pdf/ambidvd2001.pdf;
judged.
Internet; accessed on 28 March 2004.
In the past, acoustical consultants could only try to
Gibbs, Barry and Oldham, David,  building
convince the client/architect that with calculations and
Acoustics , Multi-Science 2004, [article on-line],
geometrical plots that they could create an acoustically
available from http://www.mulit-
exceptional space.
science.co.uk.lba_09.htm; Internet; accessed on 28
In the simulation of the unbuilt space discussed in this
March 2004.
paper, architects and acousticians collaborated to create
Malham, D.J.,  Spatial Hearing Mechanisms and Sound
iterations of the material treatment until the acoustic
Reproduction , University of York 1998, [article
goals were achieved. The client was able to identify the
on-line], available from
influence of sound reflections and different material
http://www.york.ac.uk/inst/mustech/3d_audio/ambi
treatement schemes and relate them to a subjective
s2.htm; Internet; accessed on 9 March 2004.
impression of the sound in the proposed space. To
Rindell, J.H.,  The use of Computer Modeling in Room
Acoustics , Journal of Vibroengineering, No.3.
2000.
SimBuild2004
I or we, the author(s) of the submitted paper, have read the following Copyright Transfer and Disclaimer,
and agree to them by submitting the submitted paper.
(1) The author(s) affirm that the paper has not been published elsewhere and, if the paper is accepted, will
not be published elsewhere prior to SimBuild 2004.
(2) If the paper is accepted, the author(s) will automatically grant to IBPSA-USA a nonexclusive, royalty-
free, perpetual, worldwide, irrevocable, sub-licensable, and transferable license to publish the paper (in
unmodified form) in any fashion (including but not limited to inclusion in the SimBuild 2004 CD
proceedings, via electronic means such as the world wide web, and inclusion in future compilations of
papers). This "nonexclusive" license means that the author(s) are not restricted as to future use of the
material except that exclusive rights cannot be granted to another.
(3) The author(s) affirm that they have the right to grant the license specified in (2), that is, publication by
IBPSA-USA or its licensees will not be in conflict with copyright or other restrictions.
(4) The author(s) acknowledge that acceptance of the paper does not imply IBPSA-USA's endorsement of or
agreement with the ideas presented. Under no circumstances shall IBPSA-USA be liable for any damages
resulting from use information included in the paper.