语音科学-SpeechScienceChangLiuTheUniversityofTexasatAustin北京语言大学,2015年12月内容提纲语音科学简介语音发音的声学模型语音(元音)声学语音(元音)感知语音科学简介语音的定义(I)什么是语音?–Spokenorvocallanguage?–语音是带有语言信息的声音;Simplyspeaking,speechisconfinedtothosesoundsthatimpartlinguisticmeaningasamodalityoflanguage.语音的定义(II)语音的主要属性–听觉和发声通道Auditory-vocalchannel–传播性Broadcasttransmissionanddirectionalreception–快速衰减Rapidfading–特殊性Specialization–分段性Discreteness语音信息(I)音位信息Phoneticquality–Linguisticcontentofaspeechlanguage感情信息Affective(emotional)quality–Paralinguistic–Contributetotheinterpretationofaspeechmessageregardingthespeaker’semotion.语音里所包含的四种信息语音信息(II)个人信息Personalquality–Extralinguistic–Informativeonthespeaker’sgender,age,andstateofhealth方向信息Transmittalquality–Informationrelativetoperspectiveonthetalker’slocation语音信息(III)语音所包含的信息远大于声音所传递的语言信息Speechismorethanalinguisticmessagecommunicatedbysound.语音的语言角色(I)音位Phonemes–语言中的最小单位Thesmallestelementandunitfromwhichallwordsareformed.–Definedbythefunctioninlanguagesystem(linguistically,insteadofacoustically)–Vowelsandconsonants语音的语言角色(II)Phonemes–Phone(音素)asoundpatternwithsomeacousticfeaturesthatarestronglysimilartootherexamplesoundsofthephoneme,e.g.[s]–Allophone(变体音素)Asystematicvariantofaphoneme:seed,steed,speed.Syllables,words,andsentences–有限的单元和无限的组合Finiteunits(phonemes)andinfinitecombinations(word/sentences)语音的语言角色(III)国际音标和字母表–DiscrepancybetweenthealphabeticrepresentationandsoundpatterninEnglish–Notone-to-onepatternOne-to-manyMany-to-oneLinguisticroleofspeech(IV)音系学和语音学Phonologyandphonetics–音系学研究的是语音的结构和功能。Phonologyisabranchoflinguisticsconcernedwithhowspeechsoundsstructureandfunctioninlanguage.–音位学研究的是语音的形成与声学特性Phoneticsisconcernedwiththewaysinwhichspeechsoundsareformedandacousticfeaturesofspeech.PhysiologicalphoneticsAcousticalphoneticsPerceptualphonetics语音学的研究范围三个领域Threearenas–Physiologicphonetics–Acousticphonetics–Perception小结–语音科学简介语音的定义语音所包含的四种信息语音的单元–音位Phoneme语音科学的范围语音发声的声学模型-AcousticTheoryofSpeech(vowel)Production语音发音的声源-通道理论Source-filtertheoryofspeechproduction声源–声门的振动:Glottalsoundsource声道的过滤(共振):Filterbythevocaltract嘴唇的开张的辐射特性:Radiationcharacteristicsofthemouth-opening语音发音的声源(I)元音的声源Vowelsoundsource–声带和声门Vocalfoldsandglottis发声Phonation–Rapidvariationofthenarrowglottisaperturetoproduceapulsingsoundsource语音发音的声源(II)Positionsofvocalfoldsforbreathingandphonation语音发音的声源-发声的机制(I)Aseriesofhigh-speedcinematographyforonecycleofglottalactiononphonation语音发音的声源-发声的机制(II)声源的波形–Glottalarea–GlottalairflowGlottalareaandariflowfunctionsforphonation声源的声学特性(I)两个特性–基频和谐波F0andtheharmonics–频谱的斜率:Theamplitudepatternofthecomponentsoverfrequency:-12dB/octaveSpectrumoftheglottalairflowwave声源的声学特性(II)声高Vocalpitch(F0)–VocalpitchdoesNOTchangethespectralshape.说话的方式Styleandforceofspeaking–PressedvoiceRelativelyhighamplitudeofhigh-frequencyharmonics.声道的共振(过滤性)-(I)声道模型Modelofthepharyngeal-oraltract–Vowelshapingbytheconfigurationofthepharyngeal-oraltract.–管道模型TubemodelVocaltractshapeandtubemodel声道的共振(过滤性)-(II)Spectrumoftheneutralvowel/ə/–Resonantfrequencies500,1500,2500Hz…Theprocessoftheproductionoftheneutralvowel/ə/声道的共振(过滤性)-共振峰(I)共振峰Formant–声道的共振ResonanceofthevocaltractF1,F2,F3,andF4…–共振峰的位置LocationofformantfrequencyLengthofthevocaltractLocationoftheconstrictionDegreeofthenarrownessoftheconstriction声道的共振(过滤性)-共振峰(II)Effectsofvocal-tractlengthonspectrumoftheneutralvowel/ə/声道的共振(过滤性)-共振峰(III)共振峰频率的简单规则Formantfrequencylocation–simplerules–F1Verticalpartofthevocaltract–F2Horizontalpartofthevocaltract–Lip-roundingruleLengtheningvocaltractreducesformantfrequency声道的共振(过滤性)-共振峰(IV)Vocaltractshapesforfrontandbackvowels嘴唇张开的影响Theadditionalfilteringeffectwhenthespeechsoundisradiatedbeyondthemonthintotheatmosphere.–高通滤波器High-passfilter:+6dB/Oct.元音发音的声学模型(I)声源和共振Sourceandfiltering–P(f)=U(f)*T(f)*R(f)–U(f)andR(f)isconsideredtobeconstantDiagrammaticrepresentationofthesource-filtertheoryforvowelproduction元音发音的声学模型(II)Examplesoftheproductionofmodelvowelsaccordingtothesource-filtertheory小结–语音发音的声学模型三个部分–发声Phonation–声道的共振ShapingofvocaltractResonanceofthevocaltract-formant–嘴唇张开的辐射效应Radiationcharacteristic元音声学-VowelAcoustics语音的声学分析(I)频谱信息Spectralinformation–Spectrum–Spectrogram基频F0–F0andf0contour时域信息Temporalpattern:onset,steady-state,offset时长Duration语音的声学分析(II)声波Waveform语谱Spectrogram频谱Spectrum/ε/vowel语音的声学分析(III)Relationshipsbetweenwavesegmentsandcorrespondingspectrograms语音的声学分析(IV)波形图和基频走势WaveformandF0contourforvowel/ε/语音的声学分析(V)元音语谱图VowelSpectrograms元音分类(I)FrontvowelsBackvowelsCentralvowels元音分类(II)Formantfrequenciesofvowels影响元音频谱的因素(I)声源Glottalsoundsource–Vocalpitch,F0.–VocaleffortRelaxedphonation:steeperspectrumForcedphonation:shallowerspectrum声道形状Vocaltractconfiguration影响元音频谱的因素(II)Relaxedandforcedphonation/a/vowelwithdifferentf0s元音共振峰的多样性(I)说话人Talkerdifference–Adults–children,male–female语境Phoneticcontextdifference–DifferentadjacentconsonantsorvowelsCrowdedVowelSpace(F1XF2)byHillenbrandetal.,1995元音共振峰的多样性(II)小结–元音声学元音的主要声学特性–Formantfrequency,duration,F0,soundlevel元音共振峰的多样性–Talkerandcontexteff