《基于DSP的语音信号识别系统的研究与实现》由会员分享，可在线阅读，更多相关《基于DSP的语音信号识别系统的研究与实现（75页珍藏版）》请在金锄头文库上搜索。
1、成都理工大学硕士学位论文基于DSP的语音信号识别系统的研究与实现姓名：陈涛申请学位级别：硕士专业：信号与信息处理指导教师：孙旭20100601摘 要 I 基于基于 DSPDSP 的语音信号识别系统的研究与实现的语音信号识别系统的研究与实现 作者简介：陈涛，男，1980 年 2 月生，师从成都理工大学孙旭副教授，2010年 6 月毕业于成都理工大学信号与信息处理专业，获得工学硕士学位。 摘摘 要要 语音识别是一种最为理想的人机通信方式，随着计算机技术的进步，语音识别技术已经逐渐应用于控制、通信、消费等行业。语音识别既是理论问题，也是一个工程化问题。 它综合多学科的理论成果， 如声学、 语音学、
2、语言学、 生理学、数字信号处理、信息工程、计算机科学、模式识别、人工智能等。语音识别技术发展到今天，已经可以满足通常的应用。但是语音识别系统在商品化的进程中还存在诸如识别速度（实时性）、系统的鲁棒性以及更高的识别率等具体问题。如何提高语音识别系统这些性能，使系统更快、更稳定的工作是目前研究的重点。怎样让语音识别推向更广泛的实用领域是目前语音识别技术研究的热点。 本文就是在这种情况下对非特定人孤立词语音识别系统做了一些探讨。 论文中首先介绍了研究课题的背景和意义。 着重从语音识别技术发展的现状和目前语音识别技术面临的挑战两个方面进行了叙述。 从而说明了此课题的研究方向。 其次，论文介绍了孤立词语
3、音识别系统的基本理论，分析了孤立词语音识别的基本工作过程。接着介绍了常用的特征参数，孤立词识别算法。并对其进行比较分析，然后根据本系统的特点选用了以（美尔频率倒谱）MFCC 为特征参数，采用经典的动态时间规整（DTW）识别算法进行模板匹配，以提高识别率。提出了去噪声的短时能量端点检测方法，并用 MATLAB7.1 进行了仿真。 硬件设计方面，在介绍 TI 公司的 TMS320C6713 DSP 芯片的基础上，阐述了以 TMS320C6713 DSP 为处理中心的孤立词语音识别系统的构成，给出了各部分的具体硬件设计。硬件系统主要由四个大模块组成：DSP 模块、语音采集与处理模块、外扩存储器模块（
4、SDRAM、FLASH ROM 和 SD 卡）、电源模块和单片机控制模块。 软件设计方面，针对孤立词识别的特点，论述了系统软件的设计过程。软件设计主要包括以下几个方面:语音信号的预处理、端点检测、特征提取、建立语音模板库、模式匹配等。在软件设计过程中还给出了语音识别各部分的流程图，并对各部分用 MATLAB7.1 进行了仿真，给出了仿真结果；论述了外挂 FLASH 的TMS320C6713 引导装载程序的设计过程。通过软件系统调试，本语音识别系统具成都理工大学硕士学位论文 II 有较高的识别率。 关键词关键词：语音识别 孤立词 非特定人 动态时间规整 DSP Abstract III Rese
5、arch and Realization of Voice Signal Recognition System Based on DSP Introduction of the author: Chen Tao, male, was born in February ,1980 whose tutor was Professor Sun Xu . He graduated from Chengdu University of Technology in Signal and Information Processing major and was granted the Master Degr
6、ee in June, 2010. Abstract Speech Recognition is the best way of the personal-machine communication.By the development of computer technology, It has been implemented into product in many areas such as control, communication and consumption Not only is speech recognition a theoretical problem, but a
7、lso an engineering problem. It integrates theoretic achievements of many disciplines, for example, acoustics, phonetics, linguistics, physiology, digital processing, information engineering, computer science,pattern recognition and artificial intelligence. Speech recognition technology developed to
8、cater for the usual application . But the speech recognition system in the commercialization process there, such as the recognition speed (real time), robustness of the system and higher rates of specific issues identified . How to improve the performance of speech recognition systems, the system fa
9、ster and more stable employment is the focus of the study.How to make the speech recognition put to practical uses is the hotspots of current research of speech recognition technology.This thesis makes some discussion on speaker-independent isolated word speech recognition system in this case. The t
10、hesis first introduces the background and significance of the subject. Focus on speech recognition technology from the status quo and the current speech recognition technology challenges two aspects of the narrative. Which describes the subject of research. Secondly, The thesis introduces the elemen
11、tary theory of isolated word speech recognition system and analyses the basic work process of isolated word speech recognition. Then it introduced the characteristic parameters and isolated word speech recognition algorithms in common use, and carrying on the comparison to the all.According to the c
12、omparison results and considering this systems characteristic,it adopts MFCC characteristic parameters and the classical dynamic time warping (DTW) 成都理工大学硕士学位论文 IV recognition algorithm to improve recognition rate.In this paper,it proposes the noise short-term energy endpoint detection method ,and u
13、sing MATLAB7.1 software simulation. As for hardware design aspect, it elaborates the system constitution of isolated word speech recognition, which taking TMS320C6713 DSP as the processing center, and gives concrete hardware design of every part. The hardware system is mainly composed of four major
14、modules: DSP module, voice acquisition and processing module, spread memory module (SDRAM and FLASH ROM and SD card), the power supply module and micro-controller module. As for software design aspect,aiming at the characteristic of isolated word recognition, it elaborats the design process of syste
15、m software. The software design is primarily includes the following areas: speech signal conditioning, activity detection, feature extraction, establish a voice template library, pattern matching, etc. In the software design processing, It also gives the flow chart of every part about speech recogni
16、tion and the simulation results of every part by using MATLAB7.1 software. It also elaborates the guidance loading procedures design process of TMS320C6713 with external FLASH. According to software system debugging, the speech recognition system has a higher recognition rate . K Keywords:eywords: Speech recognition Isolated word Speaker-independent DTW DSP 独创性声明 本人声明所呈交的学位论文是本人在导师指导下进行的研究工作及取得的研究成果。据我所知，除了文中特别加以标注和致谢的地方外，论文中不包含其他人已经发表或撰写过的研究成果， 也不包含为获得 成都理工大学 或其他教育机构的学位或证书而使
