Auditory Modelling using DSAM
1.
Auditory Models at Essex
At
If you are interested in modelling, we can help you in a number of different ways. We can supply
- Auditory modelling software
- Personal advice on how to use it
- Tutorials to help you learn
- Ready made models for you to try at home including models that we have used in recent publications
1.1 Modelling software
Most of our modelling work at
MATLAB will be familiar to you but DSAM may need some introduction.
DSAM stands for Development System for Auditory Modelling. At its heart, DSAM is a bunch of C-code modules. Each module does a different task such as model the response of the basilar membrane or analyse data to create a post stimulus time histogram. If you are a C programmer, you can download the suite of modules and then take whichever modules you like for your programs.
If you are not a C-programmer, you might like to use an application called AMS (Auditory Modelling System). This is an application that allows you to name the modules you want to use and the sequence in which they are to be called. The application then does as you request and presents the results graphically or as output files that can be analysed by other programmes. This is the most popular way to use the modelling software. It does not take long to learn the conventions of AMS and there is a complete tutorial program to take you through from A to Z (well… A to T).
The most flexible way to model is a mixture of MATLAB and AMS. First you create your basic model in AMS and then you call the model from MATLAB. You can use MATLAB to create the stimulus and send it to AMS. Then AMS returns its results directly to MATLAB where they can be analysed, stored and displayed as you like it. When you install AMS, you automatically get access to a new MATLAB function runDSAMsim which calls the model and handles all the input and output information.
The model software is organised by Lowel O’Mard and you should visit his web page to download it.
http://www.essex.ac.uk/psychology/hearinglab/dsam/index.htm
However, if you are using a PC, you can shortcut some of the complications by directly downloading the WINDOWS version of AMS at
ftp://ftp.essex.ac.uk/pub/omard/dsam/AMS118v1.exe
You will also need to install the help files from
ftp://ftp.essex.ac.uk/pub/omard/dsam/DSAMHelp2660.exe
These are zipped, self-installing executables. If you are new to DSAM, you should work through the tutorials to get started.
1.2 Personal advice
We are keen that you should be able to enjoy using our software. If you need help or advice, pleas contact me (Ray Meddis) at rmeddis@essex.ac.uk .
Before writing, please download and work through the tutorial document as a courtesy. Of course, if you are having trouble installing the software or running any of the tutorials, please write immediately. At present the DSAM software runs on PCs and Linux. Macs will be accommodated in the near future, hopefully.
I can also sometimes supply ready-made models for particular applications. Please write to ask if this is possible for your particular project.
These offers also extend to commercial operators so long as the effort is not too great.
1.3 Tutorials
A tutorial document is available for beginner users of DASM as a .pdf document. If you install DSAM, the tutorial document can be found in the folder C:\DSAM|AMS|Tutorials as a zipped file Tutorials.zip or you can read it now using Acrobat if you have it installed.
It contains
- full instructions on how to install AMS
- a description of the AMS conventions
- instructions on how to use MATLAB to call run the auditory
- examples of working models of the auditory periphery
You are strongly advised to work through
the tutorial first. Most previous users
have found it quick, helpful and easy to understand. The tutorial quickly gets
you going using real auditory models. The parameter and simulation files needed
to run the examples in the tutorial are automatically supplied with the DSAM
installation and can be found in the folder C:\DSAM\AMS\Tutorials
Enjoy!
2 Working models
A number of working models can be found in the tutorial package. It is recommended that you start with these.
- Human auditory periphery (HAP) GUI
- Cochlear Nucleus pitch model
- Forward masking model
- Temproal Integration
2.1 Human auditory periphery (HAP) GUI
This model was produced some time ago but is a nice example of what can be done using a combination of MATLAB and AMS. It is a working model of the basilar membrane response (BM) to arbitrary sound stimuli. You can record sounds as .wav files and then submit them for analysis. The basis of the model is the DRNL (dual resonance non linear filterbank) using sounds read in from .wav files.
It is a MATLAB GUI interface that allows the user to nominate a .wav file for processing using the DRNL filterbank.
You can also use it for comparing the nonlinear DRNL filterbank with the more familiar gammatone filterbank.
You can read the HAP tutorial now or download it with the complete package.
The package contains MATLAB programs and AMS models along with a tutorial document and installation guide.
It requires MATLAB version 6 or later and assumes that DSAM is already installed (see Modelling software above).
Download HAP software package from here (3.5 Mbyte).
2.2 Cochlear Nucleus pitch model
This model was used in
the publication:
Wiegrebe, L. and Meddis,
R. (2004) ‘The representation of
periodic sounds in simulated sustained chopper units of the ventral cochlear nucleus’ Journal of
the Acoustical Society of America, 115, 1207-1218 .
A full description of the model is given in an appendix to the manuscript. This appendix did not appear in the published version.
You can read the introductory notes now or get them when you download the full package
The complete package can be downloaded here. This includes the article, the appendix, a readMe file and all files needed to run both the guinea pig and human models.
Here is a screen image of the guinea pig chopper model run with a single chopper
Here is a screen image of the human model run with many channels (one chopper per channel)
2.3 Forward masking model
This model was used in the publication:
Meddis ,
R. and O’Mard, L. P. (2005). “A
computer model of the auditory nerve response to forward masking stimuli.” Journal of the Acoustical Society of America.
The article has an appendix giving a full description of the model.
You can read the introductory notes now or download it with the full package.
The complete package can be downloaded here. This includes the original article, all sim and spf files needed to run the model and the MATLAB programs used to run the simulations featured in the package.
Screen image of probability version of the forward masking model
2.4 Temporal Integration
This model was used when preparing the manuscript
Meddis, R. (manuscript) Auditory nerve first-spike latency and auditory absolute threshold: a computer model.
The article describes a model that shows temporal integration in the auditory periphery and also can simulate psychophysical temporal integration (the phenomenon where the threshold of a tone is lower when the tone is longer).
Introductory notes for the software exist although it will make sense only after reading the article.
The complete package can be downloaded here.
A screen image of the model used to study first-spike latency in 20 parallel fibers with the same BF looks like this
A second model used to emulate psychophysical threshold using a cochlear nucleus model looks like this
