基于声门激励的语音语速、音量调整方法  

Speech rate and volume modification method based on glottal pulses

作　　者：杨金霄 沈天飞[1] 滕秋霞[1] 

机构地区：[1]上海大学机电工程与自动化学院,上海200072

出　　处：《电子测量技术》2016年第2期72-75,共4页Electronic Measurement Technology

摘　　要：在小词汇量孤立词语音识别的研究中,录入的语音指令语速过快(慢)或音量过大(小)都会影响语音识别系统对于这些指令的识别率。为此,需要对这些语音指令进行调整。本文提出一种基于线性预测分析技术的语音语速和音量调整方法,该方法由已知语音信号反解出声门激励信号,对反解出的声门激励信号进行语速和音量的调整,将调整后的声门激励信号再次激励声道系统输出语音信号,得到调整后的语音信号。仿真研究结果表明,该方法具有实际可行性并且可以显著改善语音识别系统对于语速过快(慢)、音量过大(小)指令的识别率。In the research on speech recognition of small vocabulary and isolated words,the recognition rate of the words with fast（or slow）rate or high（or low）volume is degraded.Therefore,it is necessary to adjust their rate or volume.Based on the linear prediction analysis technique,the paper propose a novel adjusting methodology,whereby the speech signal to be adjusted is first transformed by the anti-solution to the glottal pulse to have their rate and volume adjusted.Through the sound channel system,these adjusted glottal pulses are transformed to the adjusted speech with proper rate and volume.Simulation results show that the method is practically feasible and able to improve the recognition rate of the speech recognition system for the words with fast（or slow）rate or high（or low）volume.

关 键 词：声门激励 线性预测分析 语速调整 音量调整 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]