基于Bark域的电子耳蜗频带划分分析和拟合研究

Frequency band division analysis and fitting based on Bark domain for cochlear implant

目前电子耳蜗主流的言语处理策略是基于滤波器组的言语处理算法,该算法通过分频带进行信号处理并把参数传递到对应的电极上。电子耳蜗滤波器组的频带划分不是等分的,而是按一定规律进行并且符合人耳听觉特性的,其中,Bark域的频带划分是重要的参考。本研究基于Bark域的电子耳蜗频带划分方法,探讨Bark域频带划分的特性并结合目前的电子耳蜗滤波器组的频带划分进行分析,进而探讨频带划分中的曲线拟合方法,为电子耳蜗滤波器组中的频带划分提供重要的方法和参数。

At present,the mainstream speech processing strategy for cochlear implants is based on the speech processing algorithm of filter banks.The signals processing by the algorithm is based on frequency band division and the parameters are transferred to the corresponding electrodes.The frequency band division of cochlear filter banks is not equal,but accordes to certain rules and conforms to the human auditory characteristics.Among them,the frequency band division based on Bark domain is an important reference.Herein the method of frequency band division based on Bark domain for cochlear implants is studied.The characteristics of frequency band division based on Bark domain are discussed,and the current frequency band division is analyzed.Then the curve fitting method in frequency band division is further discussed,which provides important methods and parameters for frequency band division of cochlear filter banks.