|
Auditory Coding of Vowels in Noise
|
|
Contents
Introduction
How is speech coded in the auditory
nerve? In this practical class you will use models of auditory function to
explore the coding of vowels and other sounds in the auditory nerve. Using
MATLAB you will investigate a series of models which are commonly used to
simulate the function of the peripheral auditory system. Once you have used the
supplied models, you can write your own MATLAB functions to implement different
coding schemes, such as Seneff's generalised synchrony detector and Ghitza's
ensemble interval histogram.
Notes
You can obtain a copy of the course notes here [pdf]
Resources
You will require two packages in addition to
the stand MATLAB installation:
- MATLAB Auditory Demonstrations (MAD)
Downloading
Get the complete MAD distribution: 4.7 Meg zip archive or
gzipped tar
file
Installation
If you are installing this on individual machines, follow
the instructions here. If,
however, you wish to install this on a network disc then:
- Unpack the MAD distribution and place it in a suitable public
directory.
- Edit file madroot.m to contain the name of the installation
directory chosen above.
- Additional MATLAB functions for auditory modelling
Downloading
Get the additional functions as a zip
archive or gzipped
tar file
Installation
- Make a directory called emaster in the directory in which you
have installed the MAD package.
- Copy the emasters_setup.m file to the
MAD installation directory.
To use these packages when the have been installed, in MATLAB, change
directory to the MAD installation directory. Type emasters at the
MATLAB prompt. You can now test the installation. For example, type slt
to load the spectrogram labelling tool. It should look like the picture on the
slt help page.