现代计算机:上半月版
現代計算機:上半月版
현대계산궤:상반월판
Modern Computer
2011年
11期
3-7
,共5页
语音识剐%语音驱动%唇形动画%声学模型%HTK
語音識剮%語音驅動%脣形動畫%聲學模型%HTK
어음식과%어음구동%진형동화%성학모형%HTK
Speech Recognition%Speech Driven%Lip Animation%Acoustics Model%HTK(Hidden Markov Models)
语音驱动唇形动画的同步是人脸动画的难点之一。首先以音节为识别单位.通过严格的声韵母建模方法,利用HTK工具包,识别得到语音文件中的音节序列与时间信息;然后利用基本唇形库和音节到唇形映射表,获得与音节序列对应的唇形序列;利用唇形序列的时间信息插值播放唇形序列,实现语音驱动的唇形动画。实验表明,该方法不仅大大减少了模型数目,而且能准确识别音节序列以及时间信息,可有效地实现语音与唇动的同步。
語音驅動脣形動畫的同步是人臉動畫的難點之一。首先以音節為識彆單位.通過嚴格的聲韻母建模方法,利用HTK工具包,識彆得到語音文件中的音節序列與時間信息;然後利用基本脣形庫和音節到脣形映射錶,穫得與音節序列對應的脣形序列;利用脣形序列的時間信息插值播放脣形序列,實現語音驅動的脣形動畫。實驗錶明,該方法不僅大大減少瞭模型數目,而且能準確識彆音節序列以及時間信息,可有效地實現語音與脣動的同步。
어음구동진형동화적동보시인검동화적난점지일。수선이음절위식별단위.통과엄격적성운모건모방법,이용HTK공구포,식별득도어음문건중적음절서렬여시간신식;연후이용기본진형고화음절도진형영사표,획득여음절서렬대응적진형서렬;이용진형서렬적시간신식삽치파방진형서렬,실현어음구동적진형동화。실험표명,해방법불부대대감소료모형수목,이차능준학식별음절서렬이급시간신식,가유효지실현어음여진동적동보。
Lip synchronization is one of the key issues in speech driven face animation system. First se- lectes as recognition unit. Through a method based on rigorous initial and final modeling, uses HTK syllable to get syllable list and time information. Then uses the lip library and syllable-lip mapping table to obtain lip sequence corresponding with syllable sequence. Finally, speech driven face animation is achieved by interpolating and playing lip sequence with time informa- tion. Experimental results show that this method not only sharply reduces the number of models, but also recognizes and plots out the syllable and time information accurately in speech files. It effectively realizes the synchronization of speech and lip animation.