基于折线削波电平的自相关函数法检测基音  

Pitch Detection Based on Autocorrelation Function of Polyline Clipping Level

作　　者：张永亮[1] 李凌[2] 张先庭[1] 

机构地区：[1]南昌航空大学电子信息工程学院,江西南昌330063 [2]南京航空航天大学自动化学院,江苏南京210016

出　　处：《电声技术》2009年第1期52-54,共3页Audio Engineering

基　　金：南昌航空大学校级科研项目(EC200604057)

摘　　要：基音信息隐含于语音信号的局部极值点处,将语音信号分帧后,通过修正由局部极值点拟合的包络线得到折线削波电平,并将削波后的信号用自相关函数法估计基音。数值实验表明,语音信号经折线削波较中心削波后可保留更多的基音信息,并能有效地减少基音检测的错误。Pitch information hides in the local extrema in speech signal, and polyline clipping level is acquired by modifying the envelop of local extrema in a frame of speech. Pitch can be detected by calculating the residual signal by the method of autocorrelation function. Numerical analysis indicates that more pitch information can be reserved adopting polyline clipping level than adopting center clipping level, and the error of pitch detecting can be reduced effectively.

关 键 词：基音检测 自相关函数 削波电平 

分 类 号：TN912[电子电信—通信与信息系统]