语音增强中小波收缩阈值算法和阈值函数研究

Threshold algorithms and functions of wavelet shrinkage in speech enhancement

小波收缩用于语音增强时,阈值算法和阈值函数是最重要的参量.为了在小波语音增强中得到可靠的选择方案,通过对这两种小波收缩参数进行分析,得到了相关数据结果,并给出SURE阈值算法和Universal阈值算法的对比结论.用常见的几种阈值函数分析软类阈值函数和硬类阈值函数,并给出了这两种参数对应选择的方法.

Threshold algorithms and functions are the most important factors of wavelet shrinkage in speech enhancement. In order to get reliable selectable schemes, the paper analyzed two kinds of wavelet shrinkage parameter, compared the SURE and Universal threshold algorithms,and analyzed several kinds of threshold function, which were categorized into two major groups：soft functions and hard functions. The corresponding selection methods of the two groups of parameter are also presented.