基于加权算法的汉语语音同步三维口型动画研究  

Research on the Chinese phonetic 3D mouth animation based on weighted control algorithm

作　　者：毕永新[1] 韩慧健[2] 周世文[3] 

机构地区：[1]山东经济学院山东省数字媒体技术重点实验室,山东济南250014 [2]山东经济学院科研处,山东省数字媒体技术重点实验室,山东济南250014 [3]山东省科技馆,山东济南250011

出　　处：《图学学报》2012年第2期98-102,共5页Journal of Graphics

基　　金：国家自然科学基金资助项目(60933008,60903109,60603077);山东省自然科学基金资助项目(ZR2009GM029,ZR2009GQ004);山东省科技发展计划资助项目(2010GJC20808-13)

摘　　要：目前,国内外对人脸口型动画的研究主要是基于英文开展,而中文人脸口型动画合成方面的研究还很少。论文结合已有的研究,综合汉语拼音发音时口型变化的规律和汉语标号的时间控制,提出了声韵加权控制算法,并对整句、整段话中的标号加以权重向量的分析,能够合成与语音同步的连续变化的三维口型动画模型。在两个连续口型动画过渡处理上,通过一种余弦函数的插值方法,合成得到两个连续口型之间的过渡口型序列,从而使合成的汉语语音同步的人脸口型动画更加平滑流畅。Presently,the researches on the action of human＇s mouth shape in the world are mainly carried out based on English words.However,such research based on Chinese words is not so much.Connected with our predecessors＇ researches,this paper puts forwards a sound weighting control algorithm with the integration of the changing law of mouth shape and the time control of Chinese label when doing Chinese pronunciation.We use this algorithm to analyze the label in a sentence and a paragraph and compose a 3-dimensional model.This model can synchronize the continuously changing mouth shape with the sounds.A transition sequence of mouth shapes between two continuous mouth shapes is composed with the help of an interpolation method using cosine function.Thus,the action of human mouth synchronized with Chinese phonetics becomes more smooth and fluent.

关 键 词：汉语语音同步 三维口型动画 加权控制算法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]