首页 | 本学科首页   官方微博 | 高级检索  
     检索      

基于独立分量分析的混合声音信号分离
引用本文:吴小培,冯焕清,周荷琴,王涛.基于独立分量分析的混合声音信号分离[J].中国科学技术大学学报,2001,31(1):68-73.
作者姓名:吴小培  冯焕清  周荷琴  王涛
作者单位:1. 中国科学技术大学电子科学与技术系 安徽大学计算智能与信号处理教育部重点实验室,合肥 230039
2. 中国科学技术大学电子科学与技术系
3. 中国科学技 术大学自动化系,合肥 230026
基金项目:安徽省自然科学基金(0043214)资助项目
摘    要:论文简要介绍了有关独立分量分析(ICA)的基本理论和算法;探讨了独立分量分析在混合声音信号分离中的应用。针对ICA输出结果排序的不定性以及在长时间记录声音信号的过程,ICA混合模型系数存在时变性等问题,提出了一种结合小波变换和独立分量分析的解决方法;试验结果表明,该方法能有效地提高运算效率并获得较好的分离效果。

关 键 词:独立分量分析  声音信号  小波变换  盲源分离  负熵判决准则  FastICA算法
文章编号:0253-2778(2001)01-0068-06

Mixed Sound Signal Separation Based on Independent Component Analysis
WU Xiao pei ,FENG Huan qing ,ZHOU He qin ,WANG Tao.Mixed Sound Signal Separation Based on Independent Component Analysis[J].Journal of University of Science and Technology of China,2001,31(1):68-73.
Authors:WU Xiao pei    FENG Huan qing  ZHOU He qin  WANG Tao
Institution:WU Xiao pei 1,3,FENG Huan qing 1,ZHOU He qin 2,WANG Tao 1
Abstract:The real observed signals are always the linear mixture of some independent signal sources. Generally, the sound signals recorded in public environment can be considered as the linear mixture signals. Separating one special sound signal from the mixed sound signals is a problem of Blind Source Separation (BSS). In this paper, a new approach to BSS, called Independent Component Analysis (ICA) , is used for the separation of mixed sound signals. The basic principle of ICA is discussed in this paper. Considering that the parameters in the linear mixed model may change (time varying) during long time recording, a new method is proposed of combining wavelet transform and ICA to solve this problem. The experiment results are given in this paper.
Keywords:independent component analysis  sound signal  wavelet transform  Blind source separation
本文献已被 CNKI 维普 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号