基于ICA的重叠语音基频提取和语音增强  

Independent Component Analysis and its Application to Overlapping Speech Pitch Extraction and Speech Enhancement

作　　者：王泽[1] 朱贻盛[1] 王自明[1] 张红煊[1] 

机构地区：[1]上海交通大学生物医学工程系,200030

出　　处：《北京生物医学工程》2001年第4期241-245,278,共6页Beijing Biomedical Engineering

基　　金：国家自然科学基金支持 (基金号 :6 98710 19)

摘　　要：语音信号是一种特征时变信号 ,基音频率提取和语音增强是两种常见语音处理要求。独立分量分析 (ICA)是一种盲信号处理方法 ,目的在于将混合在观察信号中的相互独立的源信号分离出来 ,ICA在很多领域都有广泛的应用 ,在语音信号上最为成功 ,本文将ICA用来提取重叠语音信号中的基音频率 ,提出了一个基于ICA的混叠语音基频提取系统 ,成功地解决了基频相互接近时无法提取混合信号基频的问题 ,并将它应用到有音乐背景下的语音增强方面 ,均取得了较好的结果。Independent componet analysis(ICA) is a new blind source separating method, with purpose of separating independent sources from mixed signals of them. ICA has been applied in a wide range of signal processing fields. In this paper, ICA is used to do pitch extraction and speech enhancement for overlapping speeches. Good results were obtained in both two aspects. Even in the case when the primary frequencies, contained in overlapping speeches, are very close to each other, the problem was resolved successfully.

关 键 词：独立分量分析 语音基频提取 语音增强 语音信号 生物医学工程 

分 类 号：R318.04[医药卫生—生物医学工程]