小波域音频信号降噪研究  

Audio signal denoising research based on wavelet domain

作　　者：朱俊敏[1] 张潇[2] 王旌阳[1] 吴粤北[2] 

机构地区：[1]上海交通大学计算机科学与工程系,上海200240 [2]上海音乐学院音乐工程系,上海200031

出　　处：《计算机工程与应用》2009年第6期149-154,共6页Computer Engineering and Applications

基　　金：上海市科学技术委员会重点科技攻关项目~~

摘　　要：音频降噪是音频处理前期的必要步骤,小波域降噪是其中一个重要的研究方向,为了使人们对小波域音频降噪有概括了解,在对小波域音频降噪相关文献进行分析和理解的前提下,首先介绍了小波变换及其特性,同时阐述了小波降噪的最优选取和处理准则,及对音频进行小波变换时小波基的选取原则;然后评述了用于音频降噪的阈值,多小波、小波包、模极大值和尺度相关法等5类方法及各自的发展;接着介绍在前人基础上,提出新的基于模极大值和尺度理论的降噪算法;最后探讨了小波域音频降噪的发展方向。Audio denoising is the pre-step in the audio processing,and wavelet domain denoising has been a main research area.In order to give people a summary knowledge of wavelet domain audio denoising,bascd on many literatures,this paper at- tempts to make a survey of wavelet domain audio denoising.First,it introduces wavelet transformation and the characteristics,and gives the optimal selecting and dealing criterion of wavelet domain den0ising,the basis wavelet selection.Second,describes five classical methods of audio denoising,for example,threshold,multi-wavelet,wavelet packet,modulus maximum and scale correlation, then introduces their development.Third,introduces a new algorithm based on modulus maximum and scale trading theory.At the end.the future trend of wavelet domain audio denoising is pointed out.

关 键 词：小波变换 音频降噪 多小波 模极大值 小波包 

分 类 号：TN911.72[电子电信—通信与信息系统] TB5[电子电信—信息与通信工程]