基于小波变换的汉语三字词语音基频提取方法  

Research on the pitch detection method of Chinese trisyllabic words using wavelet transform

作　　者：钟金宏[1] 杨善林[1] 魏方传[1] 

机构地区：[1]合肥工业大学管理学院,安徽合肥230009

出　　处：《合肥工业大学学报（自然科学版）》2002年第3期331-334,共4页Journal of Hefei University of Technology：Natural Science

摘　　要：小波变换具有良好的时频局部分析能力 ,可利用它检测语音信号的基频。文章将其应用于具有实际连续语音特征的三字词语音的基频提取 ,发现对大多数三字词提取的基频中含有较多的错误点 ,甚至出现一堆乱点。文章分析了其产生的原因 ,给出了基于投票策略的改进方案 ,并验证了该方案的有效性。The wavelet transform is a perfect time-frequency local analysis tool, by which the pitch of speech signals can be detected. In this paper, the wavelet transform is applied to the pitch detection of trisyllabic words with the characteristics of natural continuous speeches, and a lot of error points are found among the pitch points detected in most of trisyllabic words. The causes of this phenomenon are analyzed, an improved algorithm is proposed based on the voting strategy, and its validity is proved by experiments.

关 键 词：汉语 三字词 小波变换 基频检测 投标策略 语音信号处理 

分 类 号：TN912.3[电子电信—通信与信息系统] H11[电子电信—信息与通信工程]