改进的小波阈值去噪法及其在语音信号去噪中的应用

语音信号是一种非平稳的时变信号,利用小波对非平稳信号处理具有明显的优势。小波阈值去噪算法因其算法简单,计算量小而广泛应用于信号去噪。但是硬阈值函数易造成信号振荡,软阈值函数引入固定偏移从而造成高频信息丢失。基于上述两种方法的不足,提出了一种改进的阈值函数并且对该函数引入的可变参数进行了自适应确定。仿真结果表明该方法在可以有效去除噪声的同时,减少了信号振荡,保留了信号高频成分即保留了信号尖峰点信息,改善了语音质量。

IMPROVED WAVELET THRESHOLD DE-NOISING METHOD AND ITS APPLICATION IN SPEECH SIGNAL DENOISING

The speech signal is a non-stationary time varying signal. Wavelet transform has obvious advantages in non-stationary signal processing. Wavelet threshold denoising algorithm is widely used in signal de-noising because of its simple algorithm and a small amount of computation. However, the hard threshold function is easy to cause the signal oscillation, while the soft threshold function can result the loss of high frequency information. In this paper, an improved threshold function is proposed. Moreover, the variable parameters introduced by this function are adaptively determined. The simulation results show that the method proposed can effectively remove the noise and reduce the signal oscillation. But beyond that, it can retain the high frequency components of the signal which is the representation of the signal spike point information. Finally, it improves the quality of speech signals.