小词汇量语音识别系统的设计精选.docVIP

小词汇量语音识别系统的设计 学生：詹晶晶 指导教师：沈晓波 淮南师范学院电气信息工程系 摘 要：随着语音识别理论研究的深入和数字信号处理软、硬件技术的发展，语音识别技术应用的研究越来越受到人们的关注。人类与计算机进行交流时，最直接和方便的方式就是语言交流，语音识别成功完成人机交互新领域。虽然大字表非特定人连续语音识别技术仍有诸多难点尚未解决，基特定人小字表孤立词语音识别技术(LPC和DTW算法)已相对成熟。 本文中首先对浯音识别技术研究与应用的历史和现状做了简单的介绍，并指出了该论文的研究意义所在。论文中对语音识别系统中涉及到的基本概念及原理进行了简单的分析，之后分别针对语音识别系统中预处理阶段的端点检测算法、语音信号的特征提取算法以及模式匹配和模型训练算法进行了深入的理论研究。采用“双门限”法实现语音端点检测，在实验室环境下取得了较为理想的检测效果；系统地研究了语音识别系统各种主要特征参数的提取方法，特别分析和研究了线性预测分析系数(LPC)的提取方法和美尔频率倒谱系数(MFCC)的提取方法；针对模式匹配和模型训练问题，本文采用了基于连续概率密度隐马尔可夫模型(HMM)的语音识别模型。实验证明，在信噪比较低的实验室环境下，该模型取得了良好的识别结果。 关键词：语音识别；端点检测；特征提取 The design of system of small-vocabulary and speech recognition Student: ZhanJingJing Director teacher:ShenXiaoBo Electrical Information Engineering Department of HuaiNan Normal University Abstract With the development of Speech Recognition technique and the software and hardware for Digital Signal Processing in recent years，the research of speech recognition and control system have been more and more attended．Communicating through speech with computer is the most comfort way,so the speech recognition become all important research field of modem computer technology．Although there still exist many problems which need to be overcome such as the technique of large-vocabulary,speaker-independent，continuous speech recognition，the technique(LPC and DTW)for small-vocabulary，speaker-dependent，isolated-word speech recognition has come to the age of maturity． In the paper, I first present a simple introduction of the history and the current status of the research and application of speech recognition technique，and point out the signification about our work．First we analyse some basic concept and theory about speech recognition system．Then according to the algorithms of the end—point detection for pretreatment，feature extraction of speech recognition，pattern matching and model training，we have a deep study on them．Using the “double threshold” law we can complete the end-point detection of the pronunciation，and obtain t