台灣蛙類聲紋辨識系統之開發與研究__臺灣博碩士論文知識加值 ...

... AEE)端點偵測法用來正確的萃取蛙鳴音節，此外，並提出一種能夠有效萃取蛙鳴頻譜特徵之多段式平均頻譜法(Multi-Stage Average Spectrum, MSAS)以提升辨識效果， ... 資料載入處理中... 跳到主要內容 臺灣博碩士論文加值系統 ::: 網站導覽| 首頁| 關於本站| 聯絡我們| 國圖首頁| 常見問題| 操作說明 English |FB專頁 |Mobile 免費會員 登入| 註冊 功能切換導覽列 (188.166.176.73)您好！臺灣時間：2022/10/0509:14 字體大小：       ::: 詳目顯示 recordfocus 第1筆/ 共1筆  /1頁 論文基本資料 摘要 外文摘要 目次 參考文獻 電子全文 紙本論文 QRCode 本論文永久網址: 複製永久網址Twitter研究生:陳亞仲研究生(外文):Ya-ZhongChen論文名稱:台灣蛙類聲紋辨識系統之開發與研究論文名稱(外文):TheStudyofVoiceprintRecognitionforFormosanFrogSpecies指導教授:陳文平指導教授(外文):Wen-PingChen學位類別:碩士校院名稱:國立高雄應用科技大學系所名稱:電機工程系學門:工程學門學類:電資工程學類論文種類:學術論文畢業學年度:99語文別:中文論文頁數:82中文關鍵詞:聲紋辨識、平均能量熵值端點偵測法、多段式平均頻譜法外文關鍵詞:VoiceprintRecognition、AverageEnergyEntropyEndpointDetectionAlgorithm、Multi-StageAverageSpectrum相關次數: 被引用:1點閱:658評分:下載:76書目收藏:0 在傳統生態調查皆以耗時又費力的方式進行，所幸在資通訊技術的蓬勃發展下，野外自動錄音方式已取而代之。

然而龐大的錄製音檔卻形成人員分析上的困擾，因此專家學者紛紛以生物聲紋辨識法來解決，但現今文獻卻仍有不足之處，故本論文為增進生物聲紋的辨識效能所做之研究。

本論文首先提出平均能量熵值(AverageEnergyEntropy,AEE)端點偵測法用來正確的萃取蛙鳴音節，此外，並提出一種能夠有效萃取蛙鳴頻譜特徵之多段式平均頻譜法(Multi-StageAverageSpectrum,MSAS)以提升辨識效果，其目的為求得頻譜相似之相鄰音框的平均頻譜，如此一來便可保有頻率隨時間變化特性。

在實驗中證實，此法結合音節長度特徵分類(FeatureClassificationofSyllableLength)與平均能量熵值法後，與其它不同組合的辨識演算法比較下有著最優異的辨識能力。

Itisatime-consumingandlaboriouswaybythemanpowerinthetraditionalecologicalsurvey.Fortunatelyrecentadvanceininformationandcommunicationtechnology,itcanbereplacedbytheautomaticrecordinginthefield.Neverthelesstheanalysisofmanpowerisbecomingaveryhardworkduetotheenormousrecordingaudiofiles.Therefore,manyresearchesarefocusonvoiceprintrecognitiontoaddresstheissue,butthatstillinadequate,hencethisthesisaimedtoenhancetheperformanceforrecognitionofbiologicalvoiceprint.Thisthesisfirstproposedanovelendpointdetectionmethodisnamedaverageenergyentropy(AEE)toaccuratelyextractthefrogsyllable.Besides,anexcellenceextractionfeatureoffrogspectrumisnamedmulti-stageaveragespectrum(MSAS)isalsoproposedtoincreasetherecognitionresult.TheMSAScanobtaintheaveragespectrumintheneighborframetoretainthefrequencyfeaturechangeovertime.Furthermore,itisfoundthatthecombinationMSAS,syllablelengthwithAEEprovidesthegreatestrecognitionthantheothercombinationofrecognitionmethods. 中文摘要-----------------------------------------------------------------------------i英文摘要-----------------------------------------------------------------------------ii致謝-----------------------------------------------------------------------------iii目錄-----------------------------------------------------------------------------iv表目錄-----------------------------------------------------------------------------v圖目錄-----------------------------------------------------------------------------vi一、緒論------------------------------------------------------------------------------11.1前言與動機--------------------------------------------------------------11.2相關文獻之探討--------------------------------------------------------31.3論文架構-----------------------------------------------------------------5二、相關背景知識------------------------------------------------------------------62.1訊號前處理--------------------------------------------------------------82.1.1預強調-----------------------------------------------------------82.1.2取音框-----------------------------------------------------------102.1.3加窗--------------------------------------------------------------122.2端點偵測演算法--------------------------------------------------------152.2.1時域端點偵測法-----------------------------------------------152.2.2頻域端點偵測法-----------------------------------------------202.3特徵參數擷取-----------------------------------------------------------252.3.1頻譜參數--------------------------------------------------------252.3.2倒頻譜參數-----------------------------------------------------272.3.3描述參數--------------------------------------------------------322.4比對與辨識演算法-----------------------------------------------------352.4.1動態時軸扭曲法-----------------------------------------------352.4.2整體平均聲紋頻譜法-----------------------------------------41三、研究系統架構------------------------------------------------------------------443.1平均能量熵值端點偵測法--------------------------------------------453.2特徵擷取-----------------------------------------------------------------513.3兩階段分類演算法-----------------------------------------------------543.3.1音節長度特徵分類--------------------------------------------543.3.2多段式平均頻譜法--------------------------------------------573.3.3待測音檔辨識--------------------------------------------------62四、實驗結果------------------------------------------------------------------------644.1端點偵測實驗-----------------------------------------------------------644.2蛙鳴辨識實驗-----------------------------------------------------------68五、結論與未來展望---------------------------------------------------------------755.1結論-----------------------------------------------------------------------755.2未來展望-----------------------------------------------------------------77參考文獻-----------------------------------------------------------------------------78作者簡歷-----------------------------------------------------------------------------82 [1]J.L.Shen,J.W.hungandL.S.Lee,「RobustEntropy-basedEndpointDetectionforSpeechRecognitioninNoisyEnvironments,」InternationalConferenceonSpokenLanguageProcessing,pp.1-4,1998.[2]W.H.TuandJ.W.Hung,"StudyoftheVoiceActivityDetectionTechniquesforRobustSpeechFeatureExtraction,"Proceedingsofthe19thConferenceonComputationalLinguisticsandSpeechProcessing,Sept.2007.[3]XufangZhaoandDouglasO』Shaughnessy,「Anewhybridapproachforautomaticspeechsignalsegmentationusingsilencesignaldetection,energyconvexhull,andspectralvariation,」CanadianConferenceonElectricalandComputerEngineering,pp.145-148,May.2008.[4]A.Taylor,G.Grigg,G.WatsonandH.McCallum,「MonitoringFrogCommunities:AnApplicationofMachineLearning,」ProceedingsofEighthInnovativeApplicationsofArtificialIntelligenceConference,pp.1564-1569,1996.[5]A.Harma,「Automaticidentificationofbirdspeciesbasedonsinusoidalmodelingofsyllables,」ProceedingsofIEEEInternationalConferenceonAcoustics,Speech,andSignalProcessing,vol.5,pp.545-548,2003.[6]C.H.Lee,C.H.Chou,C.C.HanandR.Z.Hunag,「AutomaticrecognitionofanimalvocalizationsusingaveragedMFCCandlineardiscriminantanalysis,」PatternRecognitionLetters,27(2),pp.93-101,Jan.2006.[7]S.B.DavisandP.Mermelstein,「Comparisonofparametricrepresentationsformonosyllabicwordrecognitionincontinuouslyspokensentences,」IEEETransactionsonAcoustics,Speech,andSignalProcessing,28(4),pp.357-366,1980.[8]H.Tyagi,R.M.Hegde,H.A.MurthyandA.Prabhaka,「Automaticidentificationofbirdcallsusingspectralensembleaveragevoiceprints,」ProceedingsoftheThirteenthEuropeanSignalProcessingConference,Sep.2006.[9]C.L.LiandK.C.Hui,「Featurerecognitionbytemplatematching,」Computers&Graphics,24(4),pp.569-582,Aug.2000.[10]C.MyersandL.R.Rabiner,「PerformanceTradeoffsinDynamicTimeWarpingAlgorithmsforIsolatedWordRecognition,」IEEETransactionsonAcoustics,Speech,andSignalProcessing,Vol.28,No.6,pp.623-635,Dec.1980.[11]J.A.KoganandD.Margoliash,「AutomatedrecognitionofbirdsongelementsfromcontinuousrecordingsusingdynamictimewarpingandhiddenMarkovmodels:Acomparativestudy,」J.Acoust.Soc.Amer.,vol.103(4),pp.2185-2196,Apr.1998.[12]T.KomoriandS.Katagiri,S,「Applicationofageneralizedprobabilisticdescentmethodtodynamictimewarping-basedspeechrecognition,」IEEEInternationalConferenceonAcoustics,Speech,andSignalProcessing,pp.497-500,1992.[13]P.Somervuo,A.Harma,andS.Fagerlund,「ParametricRepresentationsofBirdSoundsforAutomaticSpeciesRecognition,」IEEETransactionsonAudio,Speech,andLanguageProcessing,14(6),pp.2252-2263,2006.[14]C.J.Huang,Y.J.Yang,D.X.Yang,Y.J.ChenandH.Y.Wei,「RealizationofanIntelligentFrogCallIdentificationAgent,」Proceedingsofthe2ndKESInternationalconferenceonAgentandmulti-agentsystems:technologiesandapplications,vol.4953,pp.93-102,2008.[15]C.J.Huang,Y.J.Yang,D.X.Yang,andY.J.Chen,「Frogclassificationusingmachinelearningtechniques,」ExpertSystemswithApplications,36,pp.3737-3743,2009.[16]SeppoFagerlund,「Birdspeciesrecognitionusingsupportvectormachines,」EURASIPJournalonAppliedSignalProcessing,Jan.2007.[17]A.L.McllraithandH.C.Card,「BirdsongRecognitionUsingBackpropagationandMultivariateStatistics,」IEEETrans.onSignalProcessing,vol.45,No.11,pp.2740-2748,Nov.1997[18]J.KoganandD.Margoliash,「AutomatedrecognitionofbirdsongelementsfromcontinuousrecordingsusingdynamictimewarpingandhiddenMarkovmodels:acomparativestudy,」JournaloftheAcousticalSocietyofAmerica,vol.103,no.4,pp.2187-2196,Apr.1998.[19]A.L.McIlraithandH.C.Card,「BirdsongrecognitionwithDSPandneuralnetworks,」ProceedingsofIEEEConferenceonCommunications,Power,andComputing,vol.2,pp.409-414,may.1995.[20]P.SomervuoandA.Harma,「Birdsongrecognitionbasedonhistograms,」ProceedingsofIEEEInternationalConferenceSpeech,andSignalProcessing,vol.5,pp.825-828,2004.[21]C.Dietrich,G.Palm,K.RiedeandF.Schwenker「Classificationofbioacoustictimeseriesbasedonthecombinationofglobalandlocaldecisions,」PatternRecognition,Vol.37,pp.2293-2305,Dec.2004.[22]S.E.Anderson,A.S.Dave,andD.Margoliash,「Template-basedautomarecognitionofbirdsongsyllablesfromcontinuousrecordings」,JournalofAcousticalSocietyofAmerica,vol.100,no.2,pp.1209-1219,Aug.1996.[23]ShunsukeIshimitsu,MasashiNakayamaandToshikazuYoshirni,「ConstructionoftheNoise-RobustBody-ConductedSpeechRecognitionSystem,」TheSecondInternationalConferenceonInnovativeComputing,InformationandControl,pp.123-126,Sept.2007.[24]RajeshJanakiraman,J.ChaitanyaKumarandHemaA.Murthy,「RobustSyllableSegmentationanditsApplicationtoSyllable-centricContinuousSpeechRecognition,」NationalConferenceonCommunications,pp.1-5,Jan.2010.[25]王小川，語音訊號處理，修訂二版，全華圖書股份有限公司，2009。

[26]L.Lamel,L.Labiner,A.RosenbergandJ.Wilpon"Animprovedendpointdetectorforisolatedwordrecognition,"IEEETransactionsonAcoustics,Speech,andSignalProcessing,vol.29,pp.777-785,1981.[27]YeTian,ZuoyingWangandDajinLu,"Nonspeechsegmentrejectionbasedonprosodicinformationforrobustspeechrecognition,"IEEESignalProcessingLetters,vol.9,no.11,pp.364-367,Nov.2002.[28]L.RabinerandB.H.Juang,FundamentalsofSpeechRecognition.EnglewoodCliffs,NJ:Prentice-Hall,1993.[29]S.ChatterjeandT.V.Sreenivas,「OptimumTransformDomainSplitVQ,」IEEESignalProcessingLetters,vol.15,pp.285-288,Feb.2008.[30]NationalInstrumentsLabViewWebpage[Online],Available:http://www.ni.com/labview/[31]JeffreyTravis,JimKring,喬瑞萍等譯，Labview大學實用教程(第三版)，電子工業出版社，2008  電子全文  國圖紙本論文 推文 網路書籤 推薦 評分 引用網址 轉寄                                                                                                                                                                                                                    top 相關論文 相關期刊 熱門點閱論文 1. 機房異常音辨識系統之研究   無相關期刊   1. 智慧型校園樹葉物種辨識之研究 2. 用於音樂檢索的聲紋辨識改良 3. 研究開發具實況影像與環境感知能力之LEGONXT救災機器人 4. 混合狀態估測方法應用於單通道盲訊號分離與蛙鳴辨識 5. LEGOMINDSTORMSNXT機器人平台於自動化教育之研究 6. 具雙輪詢動態頻寬分配演算法於乙太被動光纖網路之研究 7. 長距離對乙太分波多工被動光纖網路效能影響之研究 8. 聲紋分析技術外牆診斷之應用研究 9. 應用最小平方法於樹葉辨識之研究 10. 生物聲紋自動辨識 11. FPOC:在中型多頻無線網狀網路中使用四個部份重疊頻道之頻道配置策略 12. 基於支援向量機之臺灣蛙類混聲識別系統 13. 類神經網路應用於船舶聲紋辨識之研究 14. 基於聽覺原理之訊號處理應用於心血管聲紋特徵分析 15. 使用能量頻譜密度法分析扇平森林生態科學園聲景在時間上的變化     簡易查詢 | 進階查詢 | 熱門排行 | 我的研究室