Silicon Cochlea is the name that has been given to an analog circuit that is inspired by the human hearing system and that extracts the various sound frequencies in an audio signal (see for example Schaik and Liu for an introduction). If you are familiar with the Fourier transform from signal processing, this circuit has a similar function. Like the biological cochlea, the 'snail house' of the inner ear, it is a chain of consecutive low pass filters with a gradually lower cut off frequency and some resonance right around that cut-off. So the stage closest to the signal input extracts the highest sound frequencies and the one furthest away the lowest frequencies. These frequency signals are then passed on as nerve pulses to the brain which performs all the sound recognition tasks us humans are capable of.
The aim of this project is to reconstruct such a filter bank in an analog integrated circuit to decompose an audio signal into its frequency components. It is part of a bigger project for neuromorphic (computing inspired by the nervous system) sound recognition, where others would work on the neural network implementation and learning algorithms for speech recognition.