資源描述:
《基于波原子變換的語音信號去噪畢業(yè)論文》由會員上傳分享,免費在線閱讀,更多相關內容在學術論文-天天文庫。
1、Contents1.INTRODUCTION12.WAVEATOMTRANSFORM42.1DefinitionofWaveAtomTransform42.21-DCharacteristicsinWaveAtomTransform62.3TheStructureof1-Dfilter-bankinWaveAtom……………………………….…103.THRESHOLDINGMETHODFORNOISEREDUCTION173.1HardandSoftThresholdings173.2DeterminationofThreshold
2、204.EXPERIMENTSANDDISCUSSIONS……………………………………………..274.1ExperimentalConditions…………………………………………………………..274.2ResultsandDiscussions285.CONCLUSION37REFERENCE38ABSTRACT(INKOREAN)40i401.INTRODUCTIONSpeechisoneofthemostnaturalandconvenientwaysofintercommunication.Thus,speechsign
3、alprocessingtechniquesforaman-machineinterfacelikespeechrecognition,speechsynthesisandspeechcodinghavedrawnmuchattention.Asspeechprocessingmovedfromthelaboratorytothefield,ithasbecomeincreasinglyimportanttodealwiththeambientnoise.Sincethequalityandintelligibilityaredeg
4、radedbytheambientnoise,theproblemofreducingnoisecomponentsofthenoisyspeechisstillregardedasanimportantissueinthefieldofspeechresearch[1-3].Mostclassicalmethodsforhandlingambientwidebandnoisewerevariantsofanapproachbasedonsubtractinganestimateofthespectrumofthenoisefrom
5、thatofthenoisyspeech[4-7].AndthenanovelapproachfornoisereductionusingthewavelettransformhasbeenproposedbyD.L.Donoho[8].ItemploysthethresholdingtechniqueinthewaveletdomainandhasshowntohavegoodpropertiesforawideclassofsignalscorruptedbyadditivewhiteGaussiannoise.Overthep
6、asttwodecades,numeroussignalprocessingtechniqueswiththewavelettransformhavebeendevelopedtomakeuseofitssuperioritytotheconventionalFouriertransform.Recently,waveatomtransformhasbeenproposed,andhasshownits40potentialindenoisingexperimentsonbothseismicandfingerprintimages
7、[1].Waveatomwasintroducedasavariantof2Dwaveletpacketsobeyingtheparabolicscaling.Inanutshell,waveatomsinterpolateexactlybetweendirectionalwaveletsandGabor,inthesensethattheperiodofoscillationsofeachwavepacket(wavelength)islinkedtothesizeoftheessentialsupport(diameter)by
8、theparabolicscaling,i.e.,wavelength~(diameter)2[1,2].Inthisthesis,weapplythe1Dwaveatomtransformtonoisyspeechsignalsfo