文档下载 文档下载导航
设为首页 | 加入收藏
搜索 请输入内容:  
 导航当前位置: 文档下载 > 所有分类 > Music Database Retrieval Based on Spectral Similarity
侵权投诉

Music Database Retrieval Based on Spectral Similarity

We present an efficient algorithm to retrieve similar music pieces from an audio database. The algorithm tries to capture the intuitive notion of similarity perceived by human: two pieces are similar if they are fully or partially based on the same score,

fornon-matchingitems.Adynamicprogrammingap-proachisusedto ndthebestmatchingthatminimizesthisdistance.Furthermore,a“linearity ltering”stepistakentoensurethatmatchingvectorsre ectalinearscalingbasedonaconsistenttempochange.3.QueryProcessing.

Allmusic lesarepreprocessedintotheintermedi-aterepresentationofspectralvectorsdiscussedear-lier.Givenaquerysoundclip(alsoconvertedintotheintermediaterepresentation),thedatabaseismatchedagainstthequeryusingourminimum-distancematch-ingandlinearity lteringalgorithms.Thepiecesthatendupwiththehighestnumberofmatchingpointsareselectedasanswerstotheuserquery.

See[12]fordetailsandanalysisofthealgorithm.

3ExperimentsandFutureWork

Weidentify vedifferenttypesof“similar”musicpairs,withincreasinglevelsofdif culty:

TypeI:Identicaldigitalcopy

TypeII:Sameanalogsource,differentdigitalcopies,possiblywithnoise

TypeIII:Sameinstrumentalperformance,differentvo-calcomponents

TypeIV:Samescore,differentperformances(possiblyatdifferenttempo)

TypeV:Sameunderlyingmelody,differentotherwise,withpossibletransposition

Soundsamplesofeachtypecanbefoundatdoc.guandang.net yangc/musicir/.Testsareconductedonadatasetof120musicpieces,eachofsize1MB.Foreachquery,itemsfromthedatabasearerankedaccordingtothenumberof nalmatchingpointswiththequerymusic,andthetop2matchesarereturned.Foreachofthe rst4similaritytypes,retrievalaccuracyisabove90%.Type-Visthemostdif cult,andbetteralgorithmsneedtobedevelopedtohandleit.

Weareexperimentingwithindexingschemes[13]inordertogetfasterretrievalresponse.Wearealsoplanningtoaugmentthealgorithmtohandletranspositions,i.e.,pitchshifts.

References

[1]J.P.Bello,G.MontiandM.Sandler,“Techniques

forAutomaticMusicTranscription”,inInternationalSymposiumonMusicInformationRetrieval,2000.

[2]S.BlackburnandD.DeRoure,“AToolforContent

BasedNavigationofMusic”,inProc.ACMMultime-dia,1998.[3]J.C.BrownandB.Zhang,“MusicalFrequencyTrack-ingusingtheMethodsofConventionaland’Narrowed’Autocorrelation”,J.Acoust.Soc.Am.89,pp.2346-2354.1991.[4]J.Foote,“ARTHUR:RetrievingOrchestralMusicby

Long-TermStructure”,inInternationalSymposiumonMusicInformationRetrieval,2000.[5]A.Ghias,J.Logan,D.ChamberlinandB.Smith,

“QueryByHumming–MusicalInformationRetrievalinanAudioDatabase”,inProc.ACMMultimedia,1995.[6]R.J.McNab,L.A.Smith,I.H.Witten,C.L.Hender-sonandS.J.Cunningham,“Towardsthedigitalmusiclibrary:Tuneretrievalfromacousticinput”,inProc.ACMDigitalLibraries,1996.[7]E.D.Scheirer,“PulseTrackingwithaPitchTracker”,

inProc.WorkshoponApplicationsofSignalProcess-ingtoAudioandAcoustics,1997.[8]E.D.Scheirer,Music-ListeningSystems,Ph.D.disser-tation,MassachusettsInstituteofTechnology,2000.[9]A.S.Tanguiane,Arti cialPerceptionandMusic

Recognition,Springer-Verlag,1993.[10]G.TzanetakisandP.Cook,“AudioInformationRe-trieval(AIR)Tools”,inInternationalSymposiumonMusicInformationRetrieval,2000.[11]E.Wold,T.Blum,D.KeislarandJ.Wheaton,

“Content-BasedClassi cation,Searchandretrievalofaudio”,inIEEEMultimedia,3(3),1996.[12]C.Yang,“MusicDatabaseRetrievalBasedonSpec-tralSimilarity”,StanfordUniversityDatabaseGroupTechnicalReport2001-14.doc.guandang.net[13]C.Yang,“MACS:MusicAudioCharacteristicSe-quenceIndexingforSimilarityRetrieval”,inIEEEWorkshoponApplicationsofSignalProcessingtoAu-dioandAcoustics,2001.[14]C.YangandT.Lozano-P´erez,“ImageDatabaseRe-trievalwithMultiple-InstanceLearningTechniques”,

Proc.InternationalConferenceonDataEngineering,2000,pp.233-243.

上一页第2页

你可能喜欢

热门文档

相关文档

站点地图 | 文档上传 | 侵权投诉 | 手机版
本站所有资源均来自互联网,本站只负责收集和整理,均不承担任何法律责任,如有侵权等其它行为请联系我们.
文档下载 Copyright 2013 doc.guandang.net All Rights Reserved.  -  email