《基于DTW模型的语音识别》-毕业论文.docVIP

西南林学院2004届本科毕业论文 本 科 毕 业（设计）论 文 基于DTW模型的语音识别 摘要：语音识别（Speech Recognition）是让机器通过识别和理解过程把语音信号转变为相应的文本或命令的技术。在课题中，通过采用DTW（Dynamic time warping, 动态时间伸缩）算法，对实现孤立词的识别进行了初步探讨和研究，实现了在MATLAB软件环境下孤立词语的语音识别，并针对DTW的主要特点及不足做出了总结。 DTW算法基于动态规划（DP）的思想，解决了孤立词发音长短不一的模板匹配问题。文中还针对动态规划的不足提出了改进。 关键词：语音识别 DTW MATLAB 动态规划 基于DTW模型的语音识别 Voice-Identification Based on DTW Model Dan Peng (Dept. of Computer and Information Science, Southwest Forestry College, Kunming, Yunan, 650224, China) Abstract: Voice-identification is a kind of technology that is using computer to transfer the voice signal to an associated text or command by identification and understand. In this paper, DTW arithmetic is adapted to study and research the implement the identification of single-word, and Speech recognition for single-word is realized by using MATLAB. In the end, this paper gets a conclusion on the feature and the shortage of DTW. DTW arithmetic based on the method of DP has solved the problem that the voice has different time during the template matching. This paper also put forward some advises about DP. Key words: Voice-Identification DTW MATLAB DP 目 录 TOC \o 1-3 \h \z \u HYPERLINK \l _To引言 PAGEREF _To\h 1 HYPERLINK \l _To1 开发背景及可行性分析 PAGEREF _To\h 1 HYPERLINK \l _To1.1 开发背景 PAGEREF _To\h 1 HYPERLINK \l _To1.2 可行性分析 PAGEREF _To\h 2 HYPERLINK \l _To2 语音识别基础 PAGEREF _To\h 2 HYPERLINK \l _To2.1 语音识别概述 PAGEREF _To\h 2 HYPERLINK \l _To2.2 语音识别系统分类 PAGEREF _To\h 3 HYPERLINK \l _To2.3 语音识别技术介绍 PAGEREF _To\h 3 HYPERLINK \l _To2.3.1 语音识别单元的选取 PAGEREF _To\h 3 HYPERLINK \l _To2.3.2 特征参数提取技术 PAGEREF _To\h 4 HYPERLINK \l _To2.3.3 模式匹配及模型训练技术 PAGEREF _To\h 4 HYPERLINK \l _To2.4 语音识别系统的基本构成 PAGEREF _To\h 5 HYPERLINK \l _To3 Matlab 简介 PAGEREF