一种优化的VMD算法及其在语音信号去噪中的应用  

An Optimized VMD Algorithm and Its Application in Speech Signal Denoising

作　　者：李宏[1] 李定文 朱海琦 田雷 李富[2] LI Hong;LI Dingwen;ZHU Haiqi;TIAN Lei;LI Fu(School of Electrical Engineering and Information,Northeast Petroleum University,Daqing 163318,Heilongjiang Province,China;No.1 Drilling Company,Daqing Drilling Engineering Company,Daqing 163458,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]大庆钻探工程公司钻井一公司,黑龙江大庆163458

出　　处：《吉林大学学报（理学版）》2021年第5期1219-1227,共9页Journal of Jilin University:Science Edition

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

摘　　要：针对非连续、非平稳语音信号中含有噪声的问题,提出一种基于参数优化的变分模态分解去噪算法.首先,利用灰狼优化算法搜寻变分模态分解算法的最优分解参数组合分解模态数K和惩罚因子α,通过使用获得的参数组合分解语音信号以获得K个特征模态函数分量IMF;其次,利用相关系数选择有效模态分量,并用小波阈值处理无效模态分量;最后,重构小波阈值处理后的模态分量和有效模态分量以对语音信号进行去噪.实验结果表明,该算法与其他经典算法相比能有效提升信噪比,降低均方误差,提高语音信号的质量.Aiming at the problem of noise in non-continuous and non-stationary speech signals,we proposed a variational mode decomposition(VMD)denoising algorithm based on parameter optimization.Firstly,the grey wolf optimization algorithm was used to search the optimal decomposition parameter combination of the VMD algorithm:decomposition mode number K and penalty factorα.By using the combination of the obtained parameter combination to decompose the speech signal,K characteristic mode function components IMF were obtained.Secondly,the effective modal components were selected by the correlation coefficient,and the invalid modal components were processed by the wavelet threshold.Finally,the wavelet threshold processed modal component and effective modal component were reconstructed to denoise the speech signal.Experimental results show that compared with other classical algorithms,the proposed algorithm can effectively improve the signal-to-noise ratio(SNR),reduce the mean square error,and improve the quality of speech signals.

关 键 词：语音信号 灰狼算法 变分模态分解 小波阈值 相关系数 

分 类 号：TP39[自动化与计算机技术—计算机应用技术] TN912.35[自动化与计算机技术—计算机科学与技术]