一种基于小波阈值的变步长LMS语音降噪算法  

A Variable Step Size LMS Speech DenoisingAlgorithm Based on Wavelet Threshold

作　　者：刘庆强[1] 郑长敏 何红凯 吴力 LIU Qingqiang;ZHENG Changmin;HE Hongkai;WU Li(School of Electrical Engineering and Information,Northeast Petroleum University,Daqing 163318,Heilongjiang Province,China;Natural Gas Branch Training Center,Daqing Oilfield Co.,Ltd.,Daqing 163453,Heilongjiang Province,China)

机构地区：[1]东北石油大学电气信息工程学院,黑龙江大庆163318 [2]大庆油田有限责任公司天然气分公司培训中心,黑龙江大庆163453

出　　处：《吉林大学学报（理学版）》2022年第4期943-949,共7页Journal of Jilin University:Science Edition

基　　金：国家重大科技专项基金(批准号:2017ZX05019-005);黑龙江自然科学基金(批准号:LH2019F004).

摘　　要：针对最小均方误差算法在语音降噪中收敛速度慢和稳态误差较大的问题,提出一种基于小波阈值的变步长最小均方误差算法.首先,利用小波软阈值算法对语音信号进行时频分析,剔除噪声的小波系数,提高降噪精度;其次,在最小均方误差算法的基础上通过建立步长因子和误差信号之间的非线性函数关系,提出一种新的变步长最小均方误差算法,对变换后的语音信号做进一步降噪处理,提高收敛速度.实验结果表明,相比于其他单一去噪算法,该算法在信噪比和均方误差上均有较大改善.Aiming at the problems of slow convergence speed and large steady-state error of the minimum mean square error algorithm in speech noise reduction,we proposed a variable-step minimum mean square error algorithm based on wavelet threshold.Firstly,the time-frequency analysis of the speech signal was carried out by using the wavelet soft threshold algorithm,the wavelet coefficients of noise were eliminated,and the noise reduction accuracy was improved.Secondly,on the basis of the minimum mean square error algorithm,we proposed a new variable step size minimum mean square error algorithm to further denoise the transformed speech signal by establishing the nonlinear function relationship between the step size factor and the error signal.The experimental results show that compared with other single denoising algorithms,this algorithm has a great improvement in signal-to-noise ratio and mean square error.

关 键 词：信号滤波 最小均方误差算法 小波阈值 信噪比 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]