The Pole-Zero Model In The Research Of Spatial Hearing

Besides intensity, tongue and timbre, human auditory perception includes the spatiality of sound, which is called spatial hearing. Spatial hearing has been a research project related to acoustics, psychological acoustics, physiological acoustics, and signal and information processing etc. for a long time. With the rapid development of computer science and information technology, investigations on spatial hearing have received more and more attention and its application has become wider and wider. Head-related impulse response (HRIR) and head-related transfer function (HRTF) play a key role in the research of spatial hearing. In this thesis, a mathematic model of human auditory space is established by using the methods of signal and information processing to approximate the HRIRs or HRTFs and study the mechanism of spatial hearing.Firstly, the measured HRIRs are pre-processed to remove the propagation delay. Then, a pole-zero model is established by Prony algorithm, which can better approximate the HRIRs. Whereafter, genetic algorithm (GA) is employed to approximate the HRTFs. Relying on the obtained preferable evolutional genes GA can model the adjacent directions'HRTFs conveniently. Considering the spectral comparability of HRTFs, the HRTF is divided into a cascade of a common, directional-independent, transfer function (CTF) and a directional transfer function (DTF) on a decibel scale. They are, respectively, modeled by a finite-impulse-response filter. In order to extract the directional-independent information more efficiently and reduce the total number of parameters while maintaining good modeling accuracy for wave reconstruction, a common-acoustical-pole and zero model (CAPZ) is established. That is, the common acoustical poles of the model do not depend on the source directions while the zeros do.