語音合成系統(tǒng)中自動分詞技術(shù)的研究.pdf

1、哈爾濱理工大學碩士學位論文語音合成系統(tǒng)中自動分詞技術(shù)的研究姓名：陳明華申請學位級別：碩士專業(yè)：微電子學與固體電子學指導(dǎo)教師：殷景華20090301哈爾濱理工大學工學碩士學位論文ResearchontheTechnologyofAutomaticSegmentationforTextToSpeechSystemAbstractTheTTStechnologyisallimportantbranchininformationprocessi

2、ngandfindsitsgreatuseintheartificialintelligenceThecoreofTTStechnologymainlyfocuses011thetextanalysisandrhythmcontrolofvoiceTheformeristhebasicmainlyincludingspecialsymbolconversionandwordsegmentationTheaccuracyofautomat

3、icsegmentationhasgreatinfluenceonthenaturaldegreeofsubsequentmoduleoutputvoiceflowwhichdorminatesinthetextanalysissystemThemainobjectiveofthepaperistodesignandrealizeaChinesewordsegmentationsystemAfteranalyzingthemajordi

4、fficultiesappearingintheautomaticsegmentionthepurposeistoreducethedifficultyandimprovetheaccuracyThesegmentionalgorithmintegratedupgradingdictionarywithmechanicalsegmentationwasadopted，andusingforthbackmatchingsegmentati

5、ontoeliminatetheambiguityImprovementhasbeenmadeinthefollowingtwoparts：thefirstoneissegmentationdictionarywhichdividesthesingularoneintothebasiconeandthecharacteristicwordoneIntheprocessofmatchingtheintegralcharacteristic

6、dictionaryimprovesmechanicalsegmentationgreatlyandcorrectsegmentationratiosofnames，placeandquantifierandatthesametimeitreducestheambiguitycausedbythesewords，savesthetimeinprocessingambiguityandacceleratesthesegmentationT

7、hesecondistheimprovementinmechanicalsegmentation，whichrealizestwowaymatchingoftheforthandbacksegmentations，andcanselecttheforthorbackmatchinginthesegrnentatingSimultaneouslythesystemrealizesthescreellsegmentationandfiles

8、egementationComparingwi廿1theformersingularonethesystemprovidestwosegmentationpatterns，andthroughcomparisonofthesegmentationresults，theaccuracyofsegmentationgainsmuchadvantageAccordingtopersonallikesandutiUtythescreensegm