基于小波减噪的基音检测改进算法  

Improved Algorithm of Pitch Detection Based on Wavelet De-noising

作　　者：赵发 孙瑞霞 ZHAO Fa, SUN Rui- xia(School of Electrical Engineering, Anhui Polytechnic University, Wuhu Anhui 241000,Chin)

机构地区：[1]安徽工程大学电气工程学院,安徽芜湖241000

出　　处：《安徽理工大学学报（自然科学版）》2018年第2期77-82,共6页Journal of Anhui University of Science and Technology:Natural Science

基　　金：安徽工程大学青年基金资助项目(2010YQ037);安徽省高校优秀青年人才基金资助项目(2012SQRL219)

摘　　要：为了提高含噪语音信号在低信噪比情况下的基音检测准确度,以带噪语音信号为研究对象,采用小波变换对带噪语音信号进行减噪处理以提高带噪语音信号的信噪比,再利用自相关函数对减噪后的语音信号进行基音检测。利用Matlab软件对具有不同信噪比的带噪语音信号进行了基音检测改进算法仿真实验。实验结果表明,当带噪语音信号信噪比下降到0d B时,改进算法基音检测正确率略高于未进行小波减噪的基音检测算法正确率。当带噪语音信号信噪比下降到-5d B时,改进算法基音检测正确率明显高于未进行小波减噪的基音检测正确率。In order to improve the accuracy of the pitch detection of noisy speech signal under low signal-to-noise ratio（ SNR）,taking noisy speech signal as the object of study,wavelet transform was used to de-noise the noisy speech signal so as to improve the SNR of the noisy speech signal,then,the autocorrelation function was used to detect the pitch of the de-noised speech signal. Matlab software was used to simulate the pitch detection of noisy speech signals with different SNR. Experimental results show that,when the SNR of the noisy speech signal droped to 0 d B,the accuracy of pitch detection with improved algorithm was slightly higher than that without wavelet de-noising. When the SNR of the noisy speech signal droped to-5 d B,the accuracy of pitch detection with improved algorithm was obviously higher than that without wavelet de-noising.

关 键 词：小波减噪 自相关函数 基音检测 算法 

分 类 号：TN912.3[电子电信—通信与信息系统]