一种语音信号基音周期时域估计算法  

A Time Domain Estimation Algorithm for Speech Signal Pitch Period

作　　者：吴树兴[1] WU Shu-xing(Beijing Information Technology College,Beijing 100015,China)

机构地区：[1]北京信息职业技术学院

出　　处：《电脑知识与技术》2019年第8期214-216,共3页Computer Knowledge and Technology

基　　金：北京市教育委员会科技计划面上项目《汉语语音合成中影响自然度的若干关键技术研究》（编号：KM201410857001）的研究成果

摘　　要：在语音识别和语音合成中,基频周期的准确估计是一项重要的语音处理环节。自相关函数法是一种适用于低信噪比下的非常有效的基音周期估计算法,这种算法需要对信号相关后的峰值位置进行准确估计。本文提出了一种峰值点位置判别方法,它可以对时域语音信号相关后的峰值点位置进行准确判断,从而比较准确地估计出语音信号的基频周期,并且复杂性不高,易于实现。In speech recognition and speech synthesis,accurate estimation of the fundamental frequency period is an important part of speech processing.The autocorrelation function method is a very effective pitch period estimation algorithm suitable for low SNR.This algorithm needs to accurately estimate the peak position after signal correlation.In this paper,a method for judging the peak position is proposed,which can accurately judge the position of the peak point after the correlation of the time domain speech signal,so as to estimate the fundamental frequency period of the speech signal more accurately,and the complexity is not high and easy to implement.

关 键 词：语音合成 基频周期 时域 自相关函数法 峰值估计 

分 类 号：TP247[自动化与计算机技术—检测技术与自动化装置]