基于汉语文本驱动的人脸语音同步动画系统  

Face and Voice Synchronization Animation System Based on Chinese Text Drive

作　　者：杜鹏[1] 房宁[2] 赵群飞[1] 

机构地区：[1]上海交通大学自动化系系统控制与信息处理教育部重点实验室,上海200240 [2]上海交通大学国际教育学院,上海200030

出　　处：《计算机工程》2012年第13期260-262,265,共4页Computer Engineering

摘　　要：为解决动画流与语音流的同步问题,设计并实现一种人脸语音同步动画系统。将所有中文音素分为16组中文可视音素,并用输入的人脸图像合成对应的关键帧,分析输入文本得到中文可视音素序列和动画的关键帧序列,将该关键帧序列与语音流对齐,在关键帧之间插入过渡帧的同时,播放语音流和动画流,以实现人脸语音同步动画。实验结果表明,该系统能产生符合人们视觉和听觉感受的人脸语音同步动画。In this paper, a novel Chinese text drive voice synchronization facial animation system is developed to synchronize the animation- stream and the speech-stream. Chinese phonemes are divided into 16 Chinese visual phoneme categories and synthesize 16 Chinese visual phoneme key frames using the input face image. The input Chinese text is analyzed to get its Chinese visual phoneme sequence and key frame sequence for facial animation. The key frame sequence is aligned to speech-stream and some transitional frames are inserted between every two adjacent key frames The Chinese text drive voice synchronization facial animation system is implemented by playing the animation-stream and the speech-stream simultaneously. Experimental results show that the system can produce the face and voice synchronous animation which fits visual and auditory experience of people.

关 键 词：人脸动画 语音同步 中文可视音素 关键帧 过渡帧 文本驱动 

分 类 号：N945[自然科学总论—系统科学]