摘要： 提出了一种实现语音直接驱动人脸唇形动画的新方法。结合人脸唇部运动机理，建立了与唇部运动相关肌肉拉伸和下颌转动的唇形物理模型，对输入的语音信号进行分析和提取其与唇部运动相关的特征参数，并直接将其映射到唇形物理模型的控制参数上，驱动唇形动画变形，实现输入语音和唇形动画的实时同步。仿真实验结果表明，该方法有效实现了语音和唇形的实时同步，唇形动画效果更接近自然，真实感更强。且该唇形物理模型独立于人脸几何模型，可广泛应用于各类人脸唇形动画的语音驱动，具有良好的普适性和可扩展性。

关键词: 唇形动画, 唇形物理模型, 语音唇形同步

Abstract: A new approach of speech-driven facial lip synchronized animation is presented. Based on the analyzing mechanism of the facial lip motion, the physical model of muscles and mandible corresponding to lip motion is established, and the input speech is analyzed and parameterized into the feature coefficients, and that are mapped to the control coefficients of the lip model, then the lip animation is driven directly by input speech, and its real-time synchronization is realized. The results of simulations show that the input speech and facial lip animation is synchronized more precisely and efficiently in this way, and the effect realities of facial lip animation during speech are greatly improved, and due to the lip model is independent to the geometric facial model, it is also suitable for speech driven lips in the field of various facial animations.

Key words: lip animation, physical lip model, speech-lip synchronization

中图分类号: 

• TP37