基于SAPI的语音驱动口型动画方法  

Speech-Driven Lip Animation Method Based on SAPI

作　　者：杨茂巍[1] 郑伯川[2] 高春梅[1] 

机构地区：[1]四川大学计算机学院机器智能实验室,成都610064 [2]西华师范大学数学与信息学院,南充637002

出　　处：《现代计算机》2012年第7期3-7,19,共6页Modern Computer

摘　　要：语音驱动口型动画是人脸表情动画中非常关键的部分。在研究语音与口型动画同步问题的基础上,提出一种真实、自然的语音口型动画实现方法。该方法首先对输入语音进行大段分割;再通过SAPI识别出具体的汉语序列信息;然后将汉语序列转换为音节序列;最后通过音节序列到口型序列的转换得到含有口型时间信息的口型序列。在动画模块中利用该口型序列驱动3D人脸模型口型动画,取得了真实、自然的语音动画同步效果。Voice-lip animation is a crucial part in facial expression animation. Presents a real, natural voice lip animation implementation method based on the synchronization between voice and lip animation. The method firstly segments speech into pieces of small speech, and identifies Chinese sequences by using SAPI. Then translates the Chinese sequences into syllable sequences, and finally translates the syllable sequences into lip sequences with lip time information. The real and natural effects of the synchronization between voice and lip animation are abtained through driving the lip animation of the 3D facial model by the lip sequences.

关 键 词：语音驱动 SAPI 口型动画 语音分割 语音识别 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]