The Study, Based On Chinese-driven Facial Speech Animation

During recent several decades, facial animation is a challenging research field in computer graphics, which is applied widely in assistant teaching, digital entertainment, internet game, Human-Computer interaction, remote conference, medical operation, facial recognition, etc.A text-driven facial speech animation system based on MPEG-4 and Chinese linguistics is achieved by using importantly TTS(Text-to-Speech) engine and OpenGL tool. This system is consist of four parts: text analysis system; speech-synthesizing system; facial animation system and speech-animation synchronizing system. The process of this system is that on the one hand, speech-synthesizing system accepts and analyses Chinese text, then synthesizes speech and generates phoneme stream with last-time information; on the other hand, facial animation system creates realistic-looking facial model based muscle-model. Phoneme-keyframe parameters and expression-keyframe parameters is used to drive mesh vertices in order to generate facial animation. Finally, real-like facial speech animation is achieved by synchronizing speech and facial animation.Coarticulation-model created by this system solves greatly the problem of Chinese coarticulation, which is a innovation of this paper. Coarticulation describes the phenomenon that pronunciation of a phoneme is affected by its adjacent phonemes in the process of continuous talking. We partition Chinese visual phonemes as fifteen groups. Speech-synthesizing system can gain the sequence of visual phonemes with time information, solves successfully the problem that TTS engine cann't generate directly Chinese phoneme and visual phoneme, which is the second innovation of this paper. Otherwise, this system generates realistic-looking 3D facial model through adjusting model and texture-mapping technology, based on common 3D facial model. Also, this system achieves facial animation with expression through defining expression mark, expression transition, blending of expression and visual phoneme.We use only mouth to simulate speaking and expression in facial animation of this system, without considering action of other facial apparatus. So, simulating action of other facial apparatus such as eyes, tongue is a research direction.