【摘要】 该文分析了人耳耳蜗及中枢听觉神经对声信号处理的机理，主要包括耳蜗对声信号的频率分析、柯蒂氏器官毛细胞的从振动机械能到电化学能的转换、听神经的侧抑制神经网络（ＬＩＮ）和增益调整四个层次，建立各个层次上听觉的数学模型。模型的最终输出即为用于识别的语音特征。辨识实验结果也表明此听觉模型特征能较好地表征语音本质。更多还原

【Abstract】 This paper analyses the principle of speech signal processing of human cochlea and auditory nerve. The four stages of processing include cochleas frequency analysis of sound, Organ of Cortis exchanging from mechanical energy to chemical energy, lateral inhibitory network (LIN) and gain adjusting. At every stage,mathematics models are given. The output of the last model is the feature used for speech recognition (SR). The result of SR test shows that the feature extracted from this auditory model represents speech signal exactly.更多还原