摘要
小波分析具有时频局部化和多分辨率特性,而小波包分析是在小波分析的基础上对信号高频部分的更精细地分解,选取合适的小波基一直是小波包去噪分析中的关键问题。将熵函数作为选取最优小波基的评价标准,通过计算语音信号小波包分解系数的熵值来确定合适的分解方式,同时采用小波包阈值去噪算法对三种小波基进行小波包去噪仿真实验,并进行对比分析。仿真实验表明,两种熵函数选取的最优小波基都能较好地消除强噪声背景下的噪声,得到信噪比较高的语音信号。
Wavelet transform has characteristics of localized time-frequency and multi-resolution,while wavelet-packet analysis,based on wavelet analysis,is the more sophisticated decomposition on the high frequency of the signal.The selection of appropriate wavelet base is always the key issue in the wavelet packet de-noising.In this paper,with entropy function as the evaluation criteria for the best wavelet bases function,and the calculation of the wavelet packet decomposition coefficients of speech signal,the appropriate decomposition is determined.At the same time,the wavelet-packet denoising of the three wavelet bases is experimentally simulated with wavelet-packet thresholding algorithm.The simulation results show that the optimal wavelet bases selected by two entropy functions,could fairly eliminate the background noise under strong noise environment,raise the SNR of voice signal.
出处
《通信技术》
2010年第12期135-137,154,共4页
Communications Technology
基金
高等学校博士学科点专项科研基金资助项目(编号:20060732002)
甘肃省自然科学基金资助项目(编号:096RJZA084)
甘肃省教育厅研究生导师科研计划项目(编号:0814-4)
关键词
语音消噪
最优小波基
小波包分析
熵函数
speech de-noising
optimal wavelet basis
wavelet-packet analysis
entropy function
