基于FICA的盲语音信号分离方法研究  

Fast ICA based Blind Source Speech Signal Separation

作　　者：王翔[1] 范启富[1] 

机构地区：[1]上海交通大学电子信息与电气工程学院, 200240

出　　处：《微计算机信息》2009年第28期176-177,96,共3页Control & Automation

摘　　要：本文在盲信号处理的背景下介绍了独立分量分析算法的原理及特点。针对语音信号复杂多变、但源信号保持相互独立的特点,采用了改进后的FICA方法对混合语音信号进行分离,最终通过仿真实验,获得了满意的分离效果。The principle and feature of independent component analysis （ICA） was illustrated as a branch of blind source signal processing. Due to the diversity of speech signal but which can keep only comparatively independence in the source signal, an enhanced method based on Fast ICA was used to separate the mixed speeches. As the final, a good result was obtained through emulating with Matlab.

关 键 词：盲信号 独立分量分析 FICA 预处理 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]