基于CEP和LPC谱提取语音信号基音周期的方法  

Speech Signal Extraction Method of Pitch Based on CEP and LPC Spectrum

作　　者：马英[1] 石小荣[1] 李海新[1] 

机构地区：[1]青海民族学院电信系,青海西宁810000

出　　处：《现代电子技术》2009年第20期150-151,154,共3页Modern Electronics Technique

摘　　要：在语音信号分析中,只有分析出可表示语音信号本质特征的参数,才有可能利用这些参数进行高效的语音通信、语音合成和语音识别等处理。因此对语音信号采用CEP和LPC谱提取语音信号基音周期的异同进行了研究,并采用Matlab实现了仿真分析。从中可以看出,LPC谱估计基音周期的算法运算量较大,而CEP谱算法更直观,且在少部分情况下基音峰会变得更突出一些,CEP谱具有更加广阔的应用前景。In voice signal analysis, by analysing the voice signal substantive characteristicsr parameter, these parameters can be used to carry on high effective processing and voice communication, speech synthesis and speech recognition. Therefore, difference between speech signal extraction of pitch by CEP and speech signal extraction of pitch by LPC spectrum are analysed,and Matlab is used to realize the simulation analysis. LPC spectrum is used to estimate tone cycle＇s algorithm which has big calculation, but the CEP spectrum algorithm is more directviewing,and the tone summit becomes prominent, CEP spectrum has widely application prospect.

关 键 词：同态分析 线性预测分析(LPC) 基音周期 MATLAB 

分 类 号：TP274[自动化与计算机技术—检测技术与自动化装置]