摘要
结合生理声学和心理声学资料,本文提出了一个由非均匀间距带通滤波器组、检测器组和主频选取机构等三部分组成的听觉模型。它们依次表征基底膜、内毛细胞和神经纤维的特性。基于所建听觉模型并结合修正的临界带宽参数构成的语音分析系统,输入模拟了鼓膜上的声压波,输出模拟了各种神经冲动图特征。语音综合系统采用简单相加法来获取重建语音。计算机模拟实验表明,重建语音是高可懂的、自然的,证明了所建听觉模型的正确性以及临界带宽参数的修正是有意义的。
A new auditory system model based on a combination of physiological andpsychological acoustic data has been proposed. This model consists of a bank of nonuniform bandpass filters, detectors and main-frequency choosing mechanisms, they act as basilar membranes, inner hair cells and nerve fibers, respectively. Combining with the improved critical bandwidth parameters, the input to this model is analogous to the pressure at the eardrum, and the output of this model simulates various features of the firing patterns. The synthes'izer obtains the resultant speech by use of the simple adding method. Computer simulations show that the resultant speech is highly intelligible and natural. The proposed model is correct, and the improvement of the critical bandwidth parameters is effective.
基金
国家自然科学基金
关键词
听觉系统
临界带
语音分析
Auditory system
Critical bandwidth
Speech analysis/synthesis
