首页 | 本学科首页   官方微博 | 高级检索  
     检索      

基于改进阈值的小波域语音增强算法
引用本文:李倩,王让定,陈金儿.基于改进阈值的小波域语音增强算法[J].宁波大学学报(理工版),2005,18(3):344-348.
作者姓名:李倩  王让定  陈金儿
作者单位:宁波大学,信息科学与工程学院,浙江,宁波,315211
基金项目:浙江省自然科学基金(Y104144);宁波市博士基金(2005A610003).
摘    要:提出了一种基于改进阈值的小波域语音增强算法.在阈值函数中引入参数,通过调整参数以获得较优的小波系数的阈值估计,使得改进阈值介于硬阈值与软阈值之间.利用改进阈值对染噪语音的小波系数进行阈值处理,达到了语音增强的目的.实验结果表明,该方法在抑制白噪声的同时减少了语音信息的损失,较硬阈值与软阈值的小波域语音增强方法相比,增强语音的信噪比、均方误差等性能均有明显的提高.

关 键 词:小波变换  阈值函数  语音增强
文章编号:1001-5132(2005)03-0344-05
收稿时间:2005-01-12
修稿时间:2005年1月12日

Wavelet Speech Enhancement Based on Improved-Thresholding
LI Qian,WANG Rang-ding,CHEN Jin-er.Wavelet Speech Enhancement Based on Improved-Thresholding[J].Journal of Ningbo University(Natural Science and Engineering Edition),2005,18(3):344-348.
Authors:LI Qian  WANG Rang-ding  CHEN Jin-er
Institution:Faculty of Information Science and Engineering, Ningbo University, Ningbo 315211, China
Abstract:A wavelet speech enhancement algorithm based on Improved-Thresholding is presented. A parameter, which can be adjusted properly to produce the best estimations of the wavelet coefficients, is imported in the Improved-Thresolding function. The parameter enables the function to become a more general case in which it incorporates the hard and soft thresholding The experiments results show that this method can effectively enhance noisy speech infected by white Gaussian noise and reduce the loss of speech information. The enhancement of the SNR and the reduction of the MSE indicate that the performance of this method is better than the hard and soft thresholding methods.
Keywords:DWT  thresholding function  speech enhancement
本文献已被 CNKI 维普 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号