文档介绍：南京师范大学泰州学院

毕业论文

题目语音端点检测算法研究

学生姓名施容容
学号 09050311
专业计算机科学与技术
班级信科 0503
指导教师焦蓬蓬

2009 年 5 月
摘要
语音端点检测是语音分析、语音合成和语音识别中的一个必要环节。尽管语音
端点检测技术在安静的环境中已经达到了令人鼓舞的准确率,但是在实际应用时由
于噪声的引入和环境的改变通常会使系统性能显著下降。语音端点检测技术要走向
实用,就必须克服鲁棒性问题,因此低信噪比噪声环境下的语音端点检测技术的意
义非常重要。
本文首先介绍了语音端点检测技术的发展及应用,然后介绍了三种典型端点检
测算法—双门限算法、基于倒普特征的算法、基于信息熵的算法,最后介绍了开发
工具 MATLAB 的特点及其在语音处理中的应用,最后在 MATLAB 中实现了双门限算法。
关键词:语音信号;端点检测,MATLAB 等等。
I
Abstract
Speech endpoint detection is a necessary link of speech analysis,
speech synthesis and speech recognition. Although Speech endpoint
detection technology has reached a accuracy rate which is encouraging in a
quiet environment, However, it will make the system performance drop
greatly because of the induction of noise and the changed environment in
practical application . If Speech endpoint detection technology can e
practical, it is necessary to e the robustness problem. Thus, in the
low SNR environment , speech endpoint detection technology is of great
significance.
First of all, the development and application of the endpoint
detection technology was introduced in this paper , then, the three typical
endpoint detection algorithm including two-threshold algorithm, the
algorithm of based on the cestrum characteristics and the algorithm of based
on information entropy was introduced, Finally, it introduced the
characteristics of MATLAB development tools and its application in speech
processing applications, two-threshold algorithm was simulated by
MATLAB.
Key words: speech signal; Endpoint detection; MATLAB
II
南京师范大学泰州学院本科生毕业(设计)论文
目录
摘要........................................................................................................................................I
Abstract.......................................................................................................................................II
目录......................................................