基于小波變換的語音分析訓(xùn)練系統(tǒng).pdf

1、Y103S359分類號：塑2ij：23UDC：Q堡密級：公玨學(xué)號；照!!魚!⑨東南大掌碩士學(xué)位論文基于小波變換的語音分析訓(xùn)練系統(tǒng)研究生姓名：塞墻筻導(dǎo)師姓名：趙卷登熬授申請學(xué)位級別王堂殛論文提交日期窒嫂曼生曼最!呈旦學(xué)位授予單位盔畝太堂答辯委員會主席翌蜩重羔：學(xué)科專業(yè)名稱塞蟄醫(yī)坐王蕉論文答辯日期至鯉魚生曼履22自學(xué)位授予日期至鯉璺生丑一。臣一評閱人鹽土i羔塹△翌2006年6月東南大學(xué)碩士學(xué)位論文AbstractChinaiSoueofth

2、ecountrieswherethespeechdiseaseoccupiesahighproportionofthepopulationIfthosespeechdiseaseshavenotreceivedspecialspeechtrainingtheyarclikelytolosethenaturalabilityofspeechcommunicationIfdevelopingthespecialsoftwarebythemu

3、ltimediaofcomputerwecantrainandcurespeechdiseasepatientByusingspeechprocesstechnologyspeechsigrlalchangedtodifferentimagesn艟patientc月1njudgetheaccuracyratingofpronouncebytheseimagesandcorrectpronounceInthisthesisitisfⅡst

4、lysununarizcdthatsomebaseknowledgeofspeechsignalandthenitisintroducedthatthespeechanalysismethodsintimeandfrequencydomainn玲t(yī)imedomainanalysisismostsimpleandintuitionisticn璩shorttimeenergytheshorttimezerocrossingrateandth

5、eshorttimeselfcorrelationaremainanalysismethodintimedomain11圮frequencydomainanalysis∞ngetmanycharacterswhichCannotgetbythetimedomainanalysisWeCangetfrequencyinformationofspeechthroughtheanalysisofthespectrogramandspeetru

6、nlwhichgetbyusingtheshorttermFouriertransform111espectrogramconexpresstherelationshipbetweentimeandfrequencydomainanalysis，anditismainanalysismethodWeintroducethatthespectrogramCangetbyusingthewavelettransform11圮spectrog

7、ramwhichgetbyusingthewaveletWansformhasgoodfrequencyresolutioninlowfrequencyandgoodtimeresolutionillhighfrequencythismoreaccordwiththecharactersofspeechWeanalyzedalotofspeechandtheresultsshowthewavelettransformhasmoreadv

8、antageAtlastitisintroducedthattherealizationofaspeechanalyzingandtrainingsystemnesystemisbasedonwavelettransformandothertimeandfrequencydomainanalysismethod耵圮controlmenuisdevelopedinVisualC60andtheanalysisprogramisdevelo

9、pedinMauab6511lcanalysisprogramiscalledinVisual()andthedevelopingtimeisshortenedThesystemCancontroltheplayingrecordingandpausingandCandisplayfigureformsoftheinputspeechbothintimedomainandfrequencydomain玎豫usercanSetupdiff

10、erentparameterbecauseofdifferentneeds，SOtheyCangetmoreexactresultThoseinarticulatetraineescancarryoutspeechtrainingbycomparingspectrogramasvisualfeedbackKeywords：SpeechSignal；SignalProcess；TheFourierTransform；TheWaveletT