基于HHT的语音特征参数提取及其在说话人识别中的应用  

Speech feature parameter extraction based on HHT and its application in speaker recognition

作　　者：刘丽伟[1] 张瑶[1] 赵孔新[1] 钱锋[1] 

机构地区：[1]长春工业大学计算机科学与工程学院,吉林长春130012

出　　处：《长春工业大学学报》2009年第6期696-701,共6页Journal of Changchun University of Technology

基　　金：吉林省科技发展计划基金资助项目(20070332)

摘　　要：阐述了分析非线性、非平稳信号的Hilbert-Huang变换(HHT)算法。针对非平稳语音信号,提出了一种基于HHT的提取语音特征参数HHT-IF的新方法,设计了基于VQ的说话人辨认系统,分别采用HHT-IF和LPCC从不同角度进行说话人辨认实验。结果表明,特征参数HHT-IF用于说话人识别是可行的。The Hilbert-Huang transform （HHT） algorithm for nonlinear and non-stationary signal analysis is discussed. Specially to non-stationary speech signals, we offer a new method based on the HHT to extract the speech feature parameters. The speaker identification system is designed based on the VQ and the experiments are carried out at different situations with both HHT-IF and LPCC. The results show that the HHT-IF is feasible for speaker recognition.

关 键 词：Hilbert—Huang变换 特征参数 VQ 说话人识别 

分 类 号：TN912.3[电子电信—通信与信息系统]