AUDITORY IMAGE MODEL Release 8.2
Medical Research Council,
Applied Psychology Unit
Cambridge, CB2 2EF, UK
At the Applied Psychology Unit, we have developed a time-domain model
of auditory processing to simulate the auditory images produced by
complex sounds like music, speech, bird song, engines, etc.* The
Auditory Image Model (AIM) constructs its simulation of what we hear
in three stages:
1) Using an auditory filterbank, it converts the digitized sound wave
into a simulation of the basilar membrane motion (BMM) that the sound
would produce in the cochlea.
2) Using a bank of haircell simulators, it 'transduces' the BMM into a
simulation of the Neural Activity Pattern (NAP) that the sound would
produce in the auditory nerve.
3) Finally, it applies a new form of Strobed Temporal Integration
(STI) to each channel of the NAP to convert the array of NAP channels
into the models simulation of our auditory image of the sound.
The NAP includes the phase-locking information encoded by the inner
haircells because it is assumed that this information plays an
important role in auditory perception and speech perception. STI
performs the temporal integration without destroying the phase-locking
information of regular sounds -- the phase locking information that we
hear. Thus, AIM is a time-domain auditory model for studying the role
of phase locking and temporal fine-structure in auditory perception.
Sequences of auditory images can be replayed to produce cartoons of
auditory events that illustrate the dynamic response of the auditory
system to everyday sounds.
* Note that the 'auditory images' constructed by AIM are meant to
represent our initial sensations or perceptions of a sound rather than
images of past events recalled from memory.
There are now a number of ways to produce auditory images, and in
October 1995, we published a short paper describing a software package
that we use to assemble and run different forms of Auditory Image
Model. The full reference for the paper is
Patterson, R.D., Allerhand, M., and Giguere, C., (1995). "Time-domain
modelling of peripheral auditory processing: A modular architecture
and a software platform," J. Acoust. Soc. Am. 98, 1890-1894.
You can read about AIM on our WWW page
You can pick up the source code and documentation from the web page,
or by annonymous ftp from
ftp.mrc-apu.cam.ac.uk directory pub/aim
This ReadMe.First file explains how to to acquire the source code, how
to compile it, and how to get started.
There are currently two releases of AIM: aimR7 and aimR8.
aimR7 is the version described in Patterson et al., (1995). It was
released in September of 1995.
aimR8 is the latest version, released originally in August 1996 as
R8.1 and again with a few bugs fixed in May 1997 as R8.2.
The differences between R7 and R8 are described in aimR8announcement.
The changes have to do with the simulation of auditory compression
and loss of phase locking at high frequencies.
There are also a number of new facilitites including
a) a Matlab interface for AIM,
b) instructions for assembling a Meddis and Hewitt (1991) model of
c) optional power compression for the functional versions of AIM,
d) a set of silent options for postscript printing of AIM displays.
This file contains:
1. General information:
A. Disclaimer and copyright.
C. Contact addresses.
D. Ftp instructions for obtaining the software.
2. Getting started:
A. Installing the software [compilation].
B. Running the model.
C. Setting the environment variables PATH and the MANPATH.
3. Compiling the model - further details.
A. X11 libraries.
B. Alternative compilers.
C. A note on color workstations.
4. Next step after installing AIM.
1. GENERAL INFORMATION:
A. Disclaimer and Copyright
Permission to use, copy, modify, and distribute the software described in
this document without fee is hereby granted for research purposes, provided
that this copyright notice appears in all copies and in all supporting
documentation, and that the software is not redistributed for any fee
(except for a nominal shipping charge). Anyone wanting to incorporate all or
part of this software in a commercial product must obtain a license from the
Medical Research Council.
The MRC makes no representations about the suitability of the software
described in this document for any purpose. It is provided `as is' without
express or implied warranty.
The MRC disclaims all warranties with regard to this software, including all
implied warranties of merchantability and fitness. In no event shall the
MRC be liable for any special, indirect or consequential damages or any
damages whatsoever resulting from loss of use, data or profits, whether in
an action of contract, negligence or other tortious action, arising out of
or in connection with the use or performance of this software.
The AIM software was developed for Unix workstations by John
Holdsworth and Mike Allerhand of the MRC APU, under the direction of
Roy Patterson. The physiological version of AIM was developed by
Christian Giguere. The options handler is by Paul Manson. The revised
SAI module is by Jay Datta. Michael Akeroyd extended the postscript
facilites and developed the xreview routine for auditory image
The project was supported by the MRC and grants from the U.K. Defense
Research Agency, Farnborough (Research Contract 2239); the EEC Esprit
BR Porgramme, Project ACTS (3207); and the U.K. Hearing Research Trust.
C. Contact Addresses:
Roy D Patterson (for the functional version of AIM)
MRC APU, 15 Chaucer Road, Cambridge CB2 2EF, England.
Phone +44 1223 355294
Fax +44 1223 359062
Christian Giguere (for the physiological version of AIM)
University of Ottawa, Ottawa, Ontario, Canada
email: firstname.lastname@example.org email@example.com
Michael Akeroyd (for postscript printing and auxtools)
MRC IHR, University Park, Nottingham, NG7 2RD, U.K.
Minoru Tsuzaki (for the matlab interface AIMMAT)
ATR HIP Seika-cho, Soraku-gun, Kyoto, Japan
D. ftp Instructions for Obtaining the Software
Access to the APU site via ftp is by the address:
Use <Name>="anonymous" and <Password>=your email address. Once inside
the APU ftp system, cd to the AIM directory, pub/aim/. The AIM
software is in the compressed archive `aimR8.tar.Z'. The ReadMe.First
file contains the text of this document. The AIM documentation is in
the compressed archive `aimR8docs.tar.Z'. Copy all the three files.
Name (mrc-apu.cam.ac.uk:you): anonymous
Password: your email address
Details of machine and address
2. GETTING STARTED:
A. Installing the Software
It is best to make a new directory (e.g. aim) for the software. Put
aimR8.tar.Z, aimR8docs.tar.Z and ReadMe.First in the directory and
unpack the source code using the commands:
zcat aimR8.tar.Z | tar xvf -
zcat aimR8docs.tar.Z | tar xvf -
The current directory should then contain a makefile and a set of
subdirectories. Five of these contain the C source code for AIM
(filter, glib, model, stitch, and wdf). The tools directory contains C
code for ancillary routines to construct stimuli and process the
multi-channel output that AIM produces. The man directory contains
online manual pages available through MANPATH and the instruction
'manaim'. The scripts directory contains guided tours of AIM in the
form of scripts that also serve to test the installation and
illustrate a typical mode of operation. The waves directory contains
test waves demonstration waves including 'hat'. These sounds were all
sampled at 20 kHz and each sample is a 2-byte word in little-endian
order (i.e. Dec and PC order). The bytes need to be reversed for
Sun, HP and SGI machines. The bin directory contains executable routines
and links to executable routines produced by the compilation.
Compile the source code using
make TARGET=<install> <machine>
where: <machine> = sun | sgi | hp | vax | dec | sungcc | linux (For
other options try 'make -help' of see: "Compiling the model - further
The file `gen' in directory bin is the AIM program itself.
There are some auxillary tools that only compile with the gcc compiler
(saisummary, xreview, synthdramp and synthirn). If you want to try
them and have a gcc compiler you can try
make CC=gcc TARGET=<auxtools> <machine>
B. Running the model.
To verify that AIM is operational, move to the bin directory and type:
This should print general usage information on the standard output.
C. Setting Up the Environment Variables PATH and MANPATH.
The software is now ready. Before setting off, however, we recommend
a) Set your PATH to include [aim_directory]/bin
(instructions in docs/aimPaths)
b) Set your MANPATH to include [aim_directory]/man
(instructions in docs/aimPaths)
3. COMPILING THE MODEL - FURTHER DETAILS.
In the root directory, the command "make help" prints a list of the
targets and parameters for compilation. The machine-specific details
refer only to the location of the X11 libraries and header files on
the target machine.
A. X11 Libraries
The AIM graphics library is based upon X11, and the library (libX11.a) and
included header files (X11/X.h and X11/Xlib.h) are expected to be in
standard places. The command "make <machine>" assigns the paths usually used
on the particular machine.
For example, the command
assigns paths so that the following files are expected:
If the X11 library and header files are in non-standard directories,
then the path for the directories can be given to the root makefile
using the parameters X11DIR (for the directory containing the files
files X11/X.h and X11/Xlib.h) and X11LIB (for the directory containing the
library libX11.a). These parameters override the default parameters assigned
for a particular machine.
For example decstation paths could also be assigned using:-
make X11DIR=/usr/include X11LIB=/usr/lib [<machine>]
sun paths could also be assigned using:-
make X11DIR=/usr/openwin/include X11LIB=/usr/lib [<machine>]
To see what the internal defaults are for a particular machine, type:
make TARGET=help <machine>
This prints a complete list of the makefile targets, arguments, and defaults
on the standard output. The values of the arguments X11DIR and X11LIB are
the default paths associated with the specified machine.
B. Alternative compilers
The default compiler is the traditional C compiler cc.
Alternative compilers can be specified by giving the name of the compiler to
the root makefile, (assuming it exists in the current path).
For example, the GNU Project C compiler, gcc, is specified as follows,
(including it's -ansi flag, for compatability with ansi cc):
make CC=gcc CFLAGS="-O -ansi" <machine>
We find following produces code which runs significantly faster that the
traditional cc compiler:
make CC=gcc CFLAGS=-O2 <machine>
C. A Note on Color Workstations
The AIM software X11 interface is designed for monochrome screens, and
not for multiplane color screens. The software will run on a color
screen, except that when bitmaps are created they are the same depth
as the screen, since they are exact copies of the screen memory.
This also happens, for example, when the "animate" option is on, or
when the "xreview" program is used. The symptoms are a very slow
response time, and huge bitmap files.
There is a hidden option "mono=on" (short for "monochrome") which
forces the bitmap to be a single plane of the screen memory. By
default, it copies plane 1; if this does not work, the plane can be
varied with the hidden option "planemask=<integer>".
4. Next step after installing AIM.
When the installation is complete go to the 'bin' directory. It
contains the compiled programs and the user ReadMe file which is the
start point for the user documentation. bin/ReadMe provides an
overview of AIM, a set of demonstrations, an overview of the AIM
documentation, and PATH information.