復雜信道下的說話人識別.pdf

1、博士學位論文論文題目復雜信道1=的說話人識別作者姓名學科專業(yè)導師姓名完成時間王仁華教授二oo七年八月爹～上囂、一舞。攀。叁一蘭拶拜掣酮渺i中～中國科學技術(shù)人學博fj學位論義摘舞AbstractTmsthesisfocusesonthesessionvariabilitytext—independentspeakerrecognitionInthisthesisWeexplorethepossibilitiestoobtainthehig

2、hidentificationratewhilemaintainingthelittlecomputationload，Theyareexploredin：l、frontfeatureextraction；2)testalgorithmofGMM_UBM；3)SVMbasedsystem；4)sessionvariabilityreductionNovelandefficientalgorithmsareproposedIncludin

3、g：ThefixedframelengthacousticfeatureisalwaysadoptedinspeakerrecognitionThevoicedandunvoicedsoundistreatedequallyButtheunvoicedsoundismorelikeawhitenoisesignal，thevoicedsoundreflectsthemovementofthevocaltractandisaperiodi

4、csignal，SOmorespeakerinformationiscontainedinthevoicedsoundThevariableframelengthfeatureextractionprocedureisproposedFurthermorethefeatureofthevoicedsoundhasmoreweightthantheunvoicedsoundinmodeltrainingprocedurenleEERoft

5、hedynamicfeatureCallbereducedby10％againstthetraditionalfixedframelengthfeatureTheGMM_UBMisthestate—oftheartsysteminspeakerrecognitionThelog—likelihoodtestalgorithmisadoptedinthetestprocedureInthisthesis，theangleofthemode

6、ldivergenceisintroducedtoreplacetheloglikelihoodinthetestprocedureTheangleofthemodeldivergenceCanacquirealmostthesarneresultasthelog—likelihoodFurthermore，thescoresoftwosystemcarlbefused，theEERofthefusedsystemCalldecreas

7、eby12％15％againsteithersystemInrecentyearstheSVMhasmadeencouragingprogressinspeakerrecognitionThreemethodsarediscussedtoimprovetheSVMbasedsystemnTheoptimizedGMMmeansupervectorandwei【ghtsupervectoraresuggestedastheinputoft

8、heSVM111esuggestedsupervectorcanoutperformthetraditionalGMMmeallsupervectorby20％inEER21ThemodeldivergeneeandtheangleofthemodeldivergencearealsoprososedastheinputoftheSVM，andtheyCanbecombinedwiththeGLDStoimpmveperformance

9、3)Duetotheinsufficiencyofdatafromtargetspeakers，theSVMalgorithmfrequentlyencounterstheclassimbalanceproblemwhichmayintroducesevereperformancedegradationTwostrategiesareproposedtoselectimpostor’ssamplesinSVMtrainingThefir