Blind separation of noisy mixed speech signals based on wavelet transform and Independent Component Analysis
Hongyan Li, Wang Hua-kui, Baojin Xiao · 2006
The research on blind source separation is a focus in the community of signal processing and has been developed in recent years. In this contribution, we propose to independent component analysis (ICA) when the measured signals are contaminated by additive noise, a method is proposed of combining wavelet threshold de-noising and independent component analysis to separate noisy mixed speech signals. Firstly, we divide each mixed speech signal into several stationary segment and estimate the threshold of each segment. Secondly, use the threshold for each segment respectively. Thirdly, we adopted a fixed-point algorithm of FASTICA to separate the de-noising mixed speech signal. The result shows that this method may reduce the affect of noise and improve the signal-noise ratio (SNR) of separation signal, accordingly renew the original speech signals preferably