欢迎来到天天文库
浏览记录
ID:36506952
大小:1.89 MB
页数:69页
时间:2019-05-11
《孤立词语音识别技术的研究与DSP实现》由会员上传分享,免费在线阅读,更多相关内容在学术论文-天天文库。
1、西南交通大学硕士学位论文孤立词语音识别技术的研究与DSP实现姓名:魏阳申请学位级别:硕士专业:通信与信息系统指导教师:王建英20060401西南交通大学硕士研究生学位论文第1I页ABSTRACTSpeechrecognizationisoneofthebasements0fthepersonnal—mechiancOmmunication.SpeechrecognizationtechnologyhasbeenimproVedrapidlyfromthemiddleoflastcentury.Thespee
2、chrecogIlizat幻ntechnologyhasbeenappHedtothejIldustry∞n“ol,cOmmunication卜andconsumptiDnareabroadly.Therefore,tlleresearchofthespeechrecognitiOnismeaningfulandtheapplicationfore鲈oundisexPansiVe.AIgOrithm0fthesystem,thehardwaredesjgnandthesoftwaredesignwereex
3、pandedoⅡ.Theanalyticalandprocessfbundationknowledgeofthespeechsignalwereintroducedfirst.AndgofⅡrther,thefundamentaltheoryofspeechrecognitionwasdiscussedparticulady.PeculiarityandarchitectureOftheDSPTMS320VC5402wasintroduced,andthespeechrecognitjonsystemwas
4、raised.HowtodesignthecircuitbasedontheTMS320VC5402wasresearched,aIldtheconsistofthesystemandhowtheprocessOfthesystemworkswasdiscussed.AIlaloganddigitalconVertisperformedbyTI.:V320AIC23,trainingandrecognizingofthespeechsignalareperformedbyDSP,aUthe109iccont
5、rolisperfomledbymecPLDchipEPM7032AE,theresultoftherecognitionwillbeoutputfinally.ContraposjtiVely,westudjedthepanicularnessoftheChinese,theprocedureofthesoftwaredesignwasintroduced.’Imeprocedureindudeofpre-processing,startandendpojntdetecting,featurevalues
6、election,pattemmatchingandthetrainingofthefbrmwork.Startandendpointdetectionisthemostcriticalstep,wecarriedoutitwithshort—timeaverageenergy,Thefeaturevalueusedinrecognitionisshorf-timeaVerageenergy.Asystemwewilldesignshouldrecognizeaparticularspeechofapart
7、icularperson,andthetimeconsumedshouldbemoreIess,thetimedom鲥ncharacte—sticweusedonly.ThekernelalgorithmisDTw(Dynamictimewarpin曲,theaIgorithmhasbeencertifiedtobeefficient,andagoodresulthasbeenachieved.西南交通大学硕士研究生学位论文第1II页l【eywol诅s:speechrecognition;short—tim
8、eaverageenergywarping(DrW);di西talsign“processor西南交通大学硕士研究生学位论文第1页1.1课题的背景及意义第1章绪论语音是人类相互间进行通信的最自然和最方便的形式,语音通信是一种理想的人机通信方式。语音通信的研究涉及到人工智能、模式识别、数字信号处理、微机技术、语言声学、语言学和认知科学等许多学科领埘”,是一个多学科的综合性研究领域,其研究成果具有重要的学术价值和应用
此文档下载收益归作者所有