Usage of blind source separation in single channel speech enhancement algorithm
盲源分离在单通道语音增强算法中的应用

A new single channel speech enhancement algorithm was put forward. A pseudo noise source was used as the input of the multiple adaptive decorrelation (MAD) algorithm to get an improved speech signal. Moreover, Daubechies wavelet was adapted to transform the signal into wavelet domain. Proper threshold function was used to select wavelet coefficients, and then the signal into time domain was synthesized. Following the scheme provided above, we can get the speech signal with higher SNR and intelligibility.