基于小波变换的语音净化新方法

语音信号由于其时变特性，传统的小波算法虽然能够衰减语音中含有的噪声，但易造成语音的失真。丈中提出了基于小波变换的语音净化新方法，改进了阈值的选择和小波系数量化算法。仿真实验结果表明，在去除噪声和提高信噪比方面，本文方法是一种有效语音净化方法。

A New Method for Speech Purification Based on Wavelet Transform

For the time-varying property of speech, conventional wavelet algorithms could remove some noisy components from the speech, and however, it is easy to cause the distortion of speech signal. A new speech purification method based on wavelet transform is proposed, which improves the selection of threshold and the quantification of wavelet coefficient. The simulation results show that the proposed method is quite effective in removing noise and improving signal-to-noise ratio.