搜索
本科毕业设计基于MATLAB的特定人语音识别算法设计摘要语言是人类交换信息最方便、最快捷的一种方式，在高度发达的信息社会中，用数字化的方法进行语音的传送、存储、识别、合成和增强等是整个数字化通信网中最重要、最基本的组成部分之一。而在随着科技技术的发展的今天，除了人与人之间的自然语言通信之外，人与机或机器与机器之间也开始使用语言。也就是因为如此，需要涉及到语音识别技术。为了解决机器能“听懂”人类的语言，在科技如此迅猛发展的今天，语音识别技术一直受到各国科学界的关注，其对计算机发展和社会生活的重要性也日益凸显出来。在孤立字语音识别中，如语音密码锁，汽车控制等领域，都运用到了特定人语音识别技术，也就是DTW算法，相对于HMM算法，DTW算法具有简单操作。在相同环境下，两者识别效果相差不大，但是HMM算法要复杂得多，主要体现在HMM算法在训练阶段需要提供大量的语音数据，而DTW算法则不需要额外的计算。所以在特定人语音识别当中，DTW算法被广泛使用。在本次设计中，将运用到MATLAB平台来对语音信号进行处理及识别。相对于C语言而言，MATLAB平台更能给用户提供一个简单易懂的代码分析窗口。而且在个性化设计中，MATLAB可以为用户提供一个人性化界面--GUI。所以，此次设计，通过MATLAB平台建立一个GUI界面，接着对一组语音信号的输入进行预处理及端点检测，提取特征参数（MFCC），形成参考模块。然后再对一组相同的语音信号输入进行同样的操作作为测试模块，与参考模块进行DTW算法进行匹配，输出匹配后的识别结果。关键词：MATLABGUI端点检测MFCCDTWDesignofSpeechRecognitionAlgorithmBasedonSpecificMATLABCaiJingzuo(CollegeofEngineering,SouthChinaAgriculturalUniversity,Guangzhou510642,China)Abstract:Languageisawayofhumanexchangeofinformationthemostconvenient,quick,highlydevelopedintheinformationsociety,forvoicetransmission,byusingthedigitalmethodofstorage,recognition,synthesisandenhancementisoneofthemostimportantpartsofthewhole,themostbasicdigitalcommunicationnetwork.Whilewiththedevelopmentofscienceandtechnologytoday,inadditiontonaturallanguagecommunicationbetweenpeople,betweenpeopleandmachineormachineandmachinearealsostartingtousethelanguage.Itisbecauseofthis,needtoinvolvethespeechrecognitiontechnology.Inordertosolvethemachinecanhearthehumanlanguage,thetechnologyissorapiddevelopmenttoday,thespeechrecognitiontechnologyhasbeenthesubjectofscientificattentionofallcountries,theimportanceofcomputerdevelopmentandsociallifeisincreasinglyprominent.Intheisolatedwordspeechrecognition,suchasvoicepasswordlock,autocontrolfield,areappliedtothespeechrecognitiontechnology,whichisrelativetotheDTWalgorithm,HMMalgorithm,DTWalgorithmhastheadvantagesofsimpleoperation.Inthesameenvironment,boththerecognitioneffectissimilar,butHMMalgorithmismuchmorecomplex,mainlyreflectedintheHMMalgorithmneedtoprovidealargeamountofspeechdatainthetrainingphase,whiletheDTWalgorithmdoesnotneedtheextracomputation.Sointhespeaker-independentrecognition,DTWalgorithmiswidelyused.Inthisdesign,willapplytotheMATLABplatformtocarryontheprocessingandrecognitionofspeechsignal.ComparedwiththeClanguage,MATLABplatformcanprovideuserswithasimplecodeanalysiswindow.Butinthepersonalizeddesign,MATLABcanprovideahumanuserinterface--GUI.So,thedesign,theestablishmentofaGUIinterfacethroughtheMATLABplatform,andthenasetoftheinputspeechsignalpretreatment,endpointdetection,featureparameterextraction(MFCC),theformationofthereferencemodule.Thenagroupofthesamespeechsignalinputtothesameoperationasatestmodule,matchingwithreferencetoDTWalgorithmmodule,outputmatchingrecognitionresults.Keywords:DTWGUIEndpointdetectionMFCCDTWI目录1前言..........................................................................................................................................11.1语音识别的历史背景.........................................................................................................11.1.1国外研究历史及现状......................................................................................................21.1.2国内研究历史及现状......................................................................................................31.2语音识别技术的应用及研究方向.....................................................................................41.3语音识别系统的基本构成.................................................................................................52语音信号的数字模型及采集................................................................................................52.1概述.....................................................................................................................................52.2语音的发音原理.................................................................................................................52.2.1人的发声器官..................................................................................................................52.2.2语音生成..........................................................................................................................72.3语音的听觉机理.................................................................................................................82.3.1听觉器官..........................................................................................................................82.3.2耳蜗的信号处理原理....................................................................................................102.4MATLAB中的语音信号模型.........................................................................................112.4.1wavrecord函数..............................................................................................................112.4.2wavplay函数.................................................................................................................123语音信号的端点检测..........................................................................................................123.1概述................................................................................