核独立分量分析在盲源信号分离中的应用研究

Research into the Application of Kernel Independent Component Analysis in Blind Sound Signal Separation

提出利用核独立分量分析来分离混合语音信号的盲信号处理方法.介绍了基于核空间ICA的原理和基本算法,然后利用核独立分量分析算法和固定点快速分离算法分离了混合语音信号.试验结果表明:利用基于核独立分量分析的方法可以得到较为理想的分离效果.

This paper proposes a method to separate mixed speech signals and to address blind sound signals using kernel independent component analysis. What is introduced then is the principle and basic algorithm of kernel independent component analysis. Mixed speech signals are separated using kernel independent component analysis algorithm and fast fixed-point separation algorithm. The experimental results indicate that the separation using kernel independent component analysis is desirable.