基于FastICA的语音盲源分离方法
Blind separation of sound signal by using FastICA

独立分量分析（ICA）在处理盲信号分离中被广泛使用，但其收敛速度较慢.为此文章重点介绍了一种更为有效的盲源分离方法――快速独立分量分析(FastICA).文章在介绍了FastICA的基本理论和方法之后，将其应用到语音分离中.在采集了三个实际的声音信号后，将三个原始信号进行混叠，在matlab仿真环境下用FastICA方法对混叠信号进行分离，将分离结果与原始信号波形进行比对，结果说明该算法具有良好的分离效果.
The independent component analysis (ICA) is a usable method to deal with a blind signal separation problem, but which converges slowly. In this paper, a more effective algorithm of blind source separation is presented, which is the fast independent component analysis (FastICA). It introduces that the basic theory and application of FastICA in sound signal separation. Three actual speech signals are factitiously mixed, and which are separated by using the conventional FastICA. The separated results show that the FastICA has good separation efficiency