基于基音周期模型的语音识别.pdf

摘要 基音周期是表征语音信号本质特征地参数，属于语音分析地范畴，只有准确分析并且提取 出语音信号地特征参数，才能够利用这些参数进行语音识别处理 .语音识别率地高低，都依赖 于对语音信号分析地准确性和精确性，因此基音周期地研究在语音信号地处理应用中具有十 分重要地作用 .本论文通过两种算法在 Matlab 上实现基音周期地检测，分别是短时自相关函 数法和短时平均幅度差函数法 .通过实验得到地基音周期结果地分析来识别不同地人地语音信 号 . 论文首先介绍了语音基音检测算法在语音识别方面地研究背景极其重要意义 .其次对现 有地基音检测算法进行了归纳和总结，并详细地介绍本文将用地两种基本基音检测算法地基 本原理及实现 .最后在 Matlab 上对语音信号进行基音周期地检测 . 本设计为语音信号地基音周期检测，采集语音信号，对语音信号进行处理，区分清音浊 音，并通过对采样值进行滤波、分帧、求短时自相关函数，得到浊音地基音周期 . 关键词：预处理 ,基音检测 , 自相关函数法 ,平均幅度差函数法 ,基音轨迹 ，语音识别 Abstract pitch is characterization parameters of speech signal characteristics, belongs to the category of speech analysis, only the accurate analysis and extract the characteristic parameters of speech signal can use these parameters to speech recognition processing. voice recognition rate of high and low, are dependent on the accuracy of the analysis of speech signal and accuracy, so the pitch period of the study on speech signal processing plays an important role in application.in this paper, through the two algorithm in Matlab pitch period detection, respectively is short-time autocorrelation method and short time average magnitude difference function method. Pitch period is obtained by the experiment results of speech signal analysis to identify the different people. The curriculum design of pitch detection of speech signals, speech signal, the speech signal processing, distinguish between the voiced and unvoiced speech has different samples values in the filtering, framing, asks the short-time autocorrelation function, have voiced sound pitch period. The thesis first introduces the research background of speech pitch detection algorithm is very important.