The Study Of Filter Bank To Simulate The Characteristics Of The Human Basilar Membrane

Cochlea is the most importance component in human hearing system, with which human can identify various sounds in different frequency bands and intensities. Using auditory filter to simulate the process of cochlea to percept sounds will have great value in several aspects like the studying of mechanism of hearing, modeling of hearing system, speech compression and speech recognition. This paper comprehensively summarize several common auditory filters on definitions, amplitude frequency response, nonlinear of the simulation of human basilar membrane, frequency selective characteristic and filtering characteristic; Meanwhile, both advantage and disadvantage were compared. At the above basis, this paper compared the above mentioned filter and makes a comprehensive summarization.In order to make a comprehensive studying of the working mechanism of the cochlea, this paper proposes a method which has good simulation results by using Gammachirp filter bank. Based on the similar properties of Gammachirp filter bank and human basilar membrane in properties such as impulse response, the amplitude frequency and so on, particular Gammachirp filter bank of which the central frequency presents uniform distribution of logarithm was applied in the process of simulation. Various signals were applied as input in the simulation tests.The results of simulation show that: with the input signal as pure tone, compounded tone and real tone, the proposed Gammachirp filter bank in this paper have good simulation performance in several important properties such as dynamic response of human basilar membrane, selection of frequency and analysis of frequency spectrum. The proposed method has good practical value in the field of Speech compress, speech recognition and artificial cochlear hearing application.