-
摘要自20世纪来，声音识别技术得到了更为迅猛的发展，其应用领域也不断扩大。尤其是声纹识别系统的应用，已经覆盖了不少需要核查身份的场合。因此，本论文基于Labview界面和Matlab编程对声纹识别系统进行具体设计。20985
- 上一篇：微型管件液压成型机器的结构设计
- 下一篇：木板钉子缺陷检查系统设计
论文介绍了基于Labview的声音采集系统的设计方法，详细说明了声音信号采集存储的过程。根据几种常用的声纹特征，设计了MFCC特征参数提取的流程，并用Matlab编写了提取程序。另外，论文还介绍了三种声音识别模型，重点描述了矢量量化模型，并利用该模型对提取的声音特征训练后形成码本，构成样本库。最后对即时声音进行采集测试，并与样本库中的码本比较判断是否匹配。
关键词 MFCC特征值 矢量量化 特征提取 说话人识别
毕业论文设计说明书（论文）外文摘要
Title Labview-Based Voice Recognition System Design
Abstract
In recent years, the rapid development of voice recognition technology is amazing. Its application extends to every aspect of our lives. Especially, with the speaker verification system, the use for identity verification has been applied in many places. Thus, the thesis gives a specific design of the speaker recognition system.
This paper introduces the design method of sound acquisition system based on Labview software and a detailed description of how to collect and store the sound signals into a computer. This paper also gives several commonly used voiceprint characteristics, and highlight the process of MFCC parameter extraction,as well as how to write its codes by Matlab software. In addition, the paper also describes three voice recognition models, focusing on vector quantization model. It gives a detailed description of how to use the vector quantization model to train codebook of voice feature, consisting of sample library; then collect and test the new sound to see if it matches with the codebook of sample library.
Keywords MFCC eigenvalues Vector Quantization Feature Extraction Speaker Recognition
目 次
1 引言 1
1.1 课题研究的背景及意义 1
1.2 声音识别系统的具体应用 1
1.3 声音识别技术的现状 2
1.4 研究工作概述 2
2 基于Labview的声音信号采集 4
2.1 Labview概述 4
2.2 声音采集系统的设计 4
3 基于Matlab的声音特征提取 10
3.1 声音特征参数概述 10
3.2 提取特征的基本过程 10
4 识别模型 24
4.1 隐马尔可夫模型 24
4.2 高斯混合模型 25
4.3 矢量量化模型 25
结 论 36
致 谢 37
参 考 文 献 38
附 录1 39
附 录2 41
附 录3 44
1 引言
关于声音，最常见的两种研究方向是：语音识别和声纹识别。本次毕业设计的课题为后者。每个人由于发声器官构造和发声习惯的不同，发出的声音都有着自己的特征。从这些声音信号中提取出表现说话人特征的信息，进而由此判断是否为此人说话，这一过程称为说话人识别或声纹识别，也可模糊称之为声音识别。所以，它并不强调说话人声音信号中表达的含义是什么，而是从中提取出代表说话人的声音特征。