结合EMD与DWT-ACF的语音基音周期检测改进算法  

Improved Algorithm for Speech Pitch Detection Combined with EMD and DWT-ACF

作　　者：张涛 章小兵 朱明星 ZHANG Tao;ZHANG Xiaobing;ZHU Mingxing(School of Electrical and Information Engineering,Anhui University of Technology,Maanshan 243000,Anhui China)

机构地区：[1]安徽工业大学电气与信息工程学院

出　　处：《噪声与振动控制》2018年第2期173-178,192,共7页Noise and Vibration Control

基　　金：安徽工业大学重大产学研项目(RD14206003)

摘　　要：针对传统小波-自相关算法在噪声环境下检测语音的基音周期会出现偏差和漏报的情况,提出一种经验模式分解下的小波-自相关的基音周期检测改进算法。该算法首先利用经验模式分解去除含噪语音趋势项并减噪,再利用改进的小波-自相关法突出每个基音周期的峰值点,提高了基音周期检测的精度。实验结果表明,该改进方法可有效改善加噪语音在基音提取上出现的偏差误报情况以及避免部分倍频和半频错误,提高基音周期检测速率及准确率。Deviation or omission may occur in speech pitch detection in Low SNR circumstances when using traditional discrete wavelet transform-autocorrelation function algorithm.This paper proposes an improved algorithm based on empirical mode decomposition(EMD)and discrete wavelet transform-autocorrelation function(DWT-ACF).First of all,the EMD process is used to remove the noisy speech trend items and reduce the noise.Then,the improved DWT-ACF is used to highlight the peak value of each pitch period to raise the accuracy of the pitch period detection.The experimental results show that the improved method can effectively reduce the frequency doubling and half frequency errors of noisy speech in pitch extraction so that the situation of the deviation or omission is improved,and the speed and accuracy of the pitch detection is raised.

关 键 词：声学 基音周期检测 小波-自相关 经验模式分解 固有模态函数 倍频和半频 

分 类 号：TN912.3[电子电信—通信与信息系统]