basilar membrane animation

introduction | demonstration | investigate | reading | credits | downloading | home

Introduction

This demonstration shows the response of a simple basilar membrane (BM) simulation to tones, impulses and noise (and combinations thereof). The underlying BM model is a linear, passive system based on a bank of gammatone filters.

The tool

Type 'bm' to launch the demo. Now select a signal from the menu at (3). You will be presented with a dialog box asking for various parameters such as duration, level etc. The resulting signal will appear in the lower display. Press the big 'run' button (1) to start the simulation. The upper display will start to vibrate. At the same time, a cursor will move along the lower signal window to indicate the current input being processed. The left end of the BM display corresponds to low frequencies -- this will become obvious once you see the thing responding to a noise stimulus (or indeed any stimulus via the illusion of propagation from right to left). The red line stores the maximum BM displacement.

Additional signals can be added in to existing ones if the accumulate box (4) is checked. This allows the response of several tones of different frequencies/levels, or tones in noise, to be investigated.

The 'params' menu allows the parameters of the BM filterbank to be modified. Specifically, the upper and lower frequencies, and the number of filters, can be changed.

Things to investigate

1. Load an impulse. Run the simulation. Note the propagation from right (high frequency) to left (low frequency). This propagation is an illusion since the model components are quasi-independent filters in each frequency band. The higher frequency filters respond faster.
2. Load a noise signal. Observe the response. Now click the accumulate box and load a 1 kHz tone. Note the presence of the tone in the BM response.
3. Load a tone on its own and observe the location of the peak in the BM displacement. Now change the tone frequency and note the change in location of the peak displacement. Add in a further tone of a different frequency. What cues apart from peak location might be used to signal tone frequency?

Further reading

• The BM model upon which this simulation is based is the impulse invariant transform approximation to the gammatone derived in Cooke (1993) Modelling Auditory Processing & Organisation, Cambridge.

Credits etc

Produced by: Martin Cooke

Release date: October 5th 1998

Permissions: This demonstration may be used and modified freely by anyone. It may be distributed in unmodified form.