The Research Of Lip-Visualization Based On Muscle Controlling Model

Lip-Visualization corresponding to speech technology is one of the hot and difficult problems in the Visual-Voice technology. Now, there are still three problems to be solved in this technology:First, most of the researchers ignore the modeling of the tongue’s motion. Second, in coarticulation phenomenon, phoneme’s static viseme contains some rules after being affected by neighboring phonemes, but, many researchers don’t associate these rules with the geometrical characteristic of coarticulation’s viseme weight function’s curve. Third, in animation’s interpolation, the linear and trigonometric function is widely used by majority, though both of them can make a smooth animation, there are also some limitations in them, that is to say, the animation can’t emphasize the key frames because of too many intermediate transition frames.In view of above problems and the high requirements of the reality in Lip-Visualization corresponding to speech technology, this paper presents a new method about controlling the movement of the mouth and tongue’s3D (three-dimensional) model. At first, we build a tongue model according to the anatomy, and a mouth model based on meshes and textures model. Then, through analyzing the lip’s motion, we use the method of dividing the orbicularis oris into two parts to imitate the lip’s motion trail. In the part of tongue’s motion simulation, according to its trail, we divide it into some kinds of mechanical motions’ combination and use four kinds of muscles to control these motions. Finally, we realize mouth’s shape when people talking, and tongue’s movements like rotating up, moving side-to-side and so on.In order to meet the characteristics of Chinese pronunciation and the speech visualization technology’s natural and continuous condition, this thesis presents a new method to create the lip animation. One the one hand, we present a coarticulation model based on differential geometry. Through using differential geometry to analyze the influence of neighboring phonemes, this model can adequately describe the overlapping articulations phonetic process phenomenon among the phonemes. This can make the animation to be more vivid. On the other hand, we present a transition frames creating method based on piecewise parabolic interpolating to realize the lip animation’s adaptive creating. In this way, we solve the problem like:shortage of key frames’show time and difficulty of key frames’ distinguish for user.