自動化音樂情感分類問題的研究.pdf

1、浙江大學(xué)計算機科學(xué)與技術(shù)學(xué)院碩士學(xué)位論文自動化音樂情感分類問題的研究姓名：孫曉煜申請學(xué)位級別：碩士專業(yè)：計算機應(yīng)用指導(dǎo)教師：湯永川20100125浙江人學(xué)碩十學(xué)位論文AbstractAbstractMusicisapopularentertainmentforpleasureininteractandhumandigitallifeAgreatofvarietymusicisincreasingatanincrediblerateThe

2、needforclarifying，organizationandmanagementofthiskindofinformationisextensivelyexpandingTherearesomeresearchworksthathaveexploitedtheapproachestoorganizingmusicinemotioncueMusicemotionclassificationisapopulartopicinrecen

3、tyearsHowever，Duetotheemotioninformationispersonalperceptual，subjectiveandambiguous，musicaffectclassificationisachallengingproblemthroughacomputationalapproachInthispaper，weadoptanaccuratedefinitionofmusicemotionwhichhas

4、beendemonstratedbypreviousresearchIntermofthisthesis，akindofmusicemotionispresentedbytwofactors：arousalandvalenceThefocusoncollectingandlabelingmusicclipstodifferentaffecttypesdependsonmanuallabelingtestThedatabaseforexp

5、erimentsonextractingemotionfrommusicisestablishedafter400clipsaresymbolizedThemusicemotionpresentedwaysisvariousDescribingtheemotioninascalemeansisthefundamentalworksofthemusicemotionclassificationFortunately，aopensource

6、systemMARSYASisaccessibleAnewmethodisattemptedwhichisnamedInformationCellMixtureModels(ICMM)toautomatethetaskofmusicemotionclassificationThisframeworkhaspotentialapplicationinbothunsupervisedconceptlearningandsupervisedc

7、lassificationlearningThisframeworkisacceptableformusicmoodclassificationbecauseemotionisavagueconceptandhasacognitivestructureTheapplicationofICMMiSalsosuitableformusicemotionclassificationKewvordsmusicemotion，classifica