声道的调频-调幅模型及其在语音分析中的应用  

AN AM-FM MODULATION MODEL OF VOCAL TRACT AND ITS APPLICATION IN SPEECH ANALYSIS

作　　者：张 磊[1] 韩纪庆[1] 王承发[1] 

机构地区：[1]哈尔滨工业大学计算机科学与技术学院

出　　处：《计算机研究与发展》2002年第6期689-695,共7页Journal of Computer Research and Development

基　　金：本课题得到国家自然科学基金(60085001);哈尔滨工业大学跨学科交叉性研究项目基金(HIT.MD200001)资助

摘　　要：传统的线性声学理论是基于这样一个假设:流经声带的气流在声道中是以平面波的形式传播.而根据Teager的研究结果,由于在声道中有涡流分布,这个假设将不成立.基于这种非线性现象的存在,Maragos提出了一个调频-调幅模型来表示语音产生的过程;近年来这个模型已经被成功地应用到语音处理的许多领域中.介绍了调频-调幅模型的产生背景及其主要理论核心,重点讨论它在语音分析中的应用及在变异语音识别中的应用前景.Traditional linear acoustic theory assumes that airflow from the vocal folds propagates through the vocal tract as a plane wave. According to Teager's studies, however, this assumption may not hold true since the vortices are distributed throughout the vocal tract. Based on the nonlinear phenomena, a more popular AM-FM modulation model is proposed by Maragos and it has been successfully used in many fields of speech processing. In this paper, the background of the AM-FM modulation model is firstly introduced, and then its theory is explained. Finally, the applications of the AM-FM model and especially its potential prospect in the stressed speech analysis are discussed in detail.

关 键 词：声道 调频 调幅模型 语音分析 能量分离算法 TEAGER能量算子 语音信号处理 语音识别 

分 类 号：TN912.3[电子电信—通信与信息系统]