分析了信号和噪声在子波域的不同特征表现,提出了一种改进的子波域语音去噪方法,该方法用一个随尺度变化的门限阈值来抑止带噪语音信号在不同尺度上噪声的子波系数,从而更好地实现在重构语音信号中消除噪声的目的。%According to the different characteristics between signal and noise on wavelet transform domain we present a modified mothed of speech denoising. In order to reduce noise in reconstructing speech signal we use a changing threshold to remove noise components from the different scale coefficients of noisy speech.
展开▼