基于矢量量化说话人识别系统的研究及实现论文.pdf

摘要语音是最简便、人们使用最频繁的交流工具,也是我们获取信息的主要途径之一。说话人识别是利用语音信号对说话人身份进行识别的一门技术,是语音信号处理领域的一个重要分支,该技术在安全保障、公安司法等领域有着广泛应用。本文以矢量量化模板训练方法为基础,建立了一套说话人识别软件系统。主要工作有: (1)详细阐述了语音信号前期处理及特征提取的方法。研究了矩形窗和海明窗对语音信号分析的影响,比较了双门限法、谱减法和谱熵法三种端点检测的方法,介绍了LPC、LSP、、、A 和A 六种特征参数的提取方法,并详细阐述了矢量量化的基本原理、过程和算法。(2)以Matlab GUI、M语言和编译器为工具,搭建了说话人识别软件系统。在系统前端,设置了登录和语音库管理模块,给用户提供了安全、简便的使用环境。在系统的主界面,设置了模式选择模块和说话人识别的各步骤模块,后者包含了之前理论部分的所有方法, 体现出的特色是:可选择方案种类多、可自行设置参数值、单个人的识别和多人识别都可进行、操作简单、效果直观。在主界面中,还设置了反馈信息模块、语音的时域、变换域分析等其他功能模块,扩大了系统的使用范围,增加了其灵活性。(3)以搭建的软件系统为平台,进行系统测试和实验分析。测试表明:系统各模块相互协调良好、能够顺利完成识别的任务;进行的三项实验分别是特征参数及其阶数、训练模型码本大小以及帧长和帧移大小对识别结果的影响,实验表明:在本实验条件下,14阶的特征参数,在训练模型码本大小为128、语音信号分帧的帧长和帧移为440和220时,识别效果最优。关键词:说话人识别,矢量量化,特征参数,软件系统 ABSTRACT Speech i s not only the most frequent ly used and the most convenient tool munication,but also the main way for US to get access to speaker recognition which is an important branch in the field of speech signal processing, is atechnique to identify the speaker by speech technique ismainly used inthe field ofsecurity,judicature and SO on. This paper builds a set of speech recognition software system based on a template training method of vector main contents are as follows: (1)Different methods of the speech signal pre—processing and feature extract ion are elaborated in thi s effects oftherectangular window andhamming window on speech signal analysis are methods of endpoint detection including double threshold method, spectral subtract ion method and spectral entropy method pared. Extract ion methods ofLPC,LSP,,,△ and are basic principles,process and algorithms of vector quantization are elaborated. (2)With Matlab GUI,M language pi ler for tools,the speech recogni tion software system i s set front of the system,the login module and speech database management module are set up to provide asafe and convenient using environment. In the main interface ofthe system,there are mode select ion mod