基于波原子变换的语音信号去噪毕业论文

《基于波原子变换的语音信号去噪毕业论文》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、Contents1.INTRODUCTION12.WAVEATOMTRANSFORM42.1DefinitionofWaveAtomTransform42.21-DCharacteristicsinWaveAtomTransform62.3TheStructureof1-Dfilter-bankinWaveAtom……………………………….…103.THRESHOLDINGMETHODFORNOISEREDUCTION173.1HardandSoftThresholdings173.2DeterminationofThreshold204.

2、EXPERIMENTSANDDISCUSSIONS……………………………………………..274.1ExperimentalConditions…………………………………………………………..274.2ResultsandDiscussions285.CONCLUSION37REFERENCE38ABSTRACT(INKOREAN)40i401.INTRODUCTIONSpeechisoneofthemostnaturalandconvenientwaysofintercommunication.Thus,speechsignalproces

3、singtechniquesforaman-machineinterfacelikespeechrecognition,speechsynthesisandspeechcodinghavedrawnmuchattention.Asspeechprocessingmovedfromthelaboratorytothefield,ithasbecomeincreasinglyimportanttodealwiththeambientnoise.Sincethequalityandintelligibilityaredegradedbytheam

4、bientnoise,theproblemofreducingnoisecomponentsofthenoisyspeechisstillregardedasanimportantissueinthefieldofspeechresearch[1-3].Mostclassicalmethodsforhandlingambientwidebandnoisewerevariantsofanapproachbasedonsubtractinganestimateofthespectrumofthenoisefromthatofthenoisysp

5、eech[4-7].AndthenanovelapproachfornoisereductionusingthewavelettransformhasbeenproposedbyD.L.Donoho[8].ItemploysthethresholdingtechniqueinthewaveletdomainandhasshowntohavegoodpropertiesforawideclassofsignalscorruptedbyadditivewhiteGaussiannoise.Overthepasttwodecades,numero

6、ussignalprocessingtechniqueswiththewavelettransformhavebeendevelopedtomakeuseofitssuperioritytotheconventionalFouriertransform.Recently,waveatomtransformhasbeenproposed,andhasshownits40potentialindenoisingexperimentsonbothseismicandfingerprintimages[1].Waveatomwasintroduce

7、dasavariantof2Dwaveletpacketsobeyingtheparabolicscaling.Inanutshell,waveatomsinterpolateexactlybetweendirectionalwaveletsandGabor,inthesensethattheperiodofoscillationsofeachwavepacket(wavelength)islinkedtothesizeoftheessentialsupport(diameter)bytheparabolicscaling,i.e.,wav

8、elength~(diameter)2[1,2].Inthisthesis,weapplythe1Dwaveatomtransformtonoisyspeechsignalsfo