Speech and expression: a computer solution to face animation

Animation which uses three dimensional computer graphics relies heavily on geometric transformations over time for the motions of camera and objects. To make a figure walk or make a liquid bubble requires sophisticated motion control not usually available in commercial animation system,. This paper describes a way to animate a model of the human face . The animator can build a sequence of facial movements, including speech, by manipulating a sma.1I set of keywords . Synchronized speech is possible because the same encoding of the phonetic elements (segments) is used to drive both the animation and a speech generator. Any facial expression , or string of segments, may be given a name and used as a key element. The final animated sequence is then generated automatically by expansion (if necessary) followed by interpolation between the resulting key frames. We present two alternative modelling techniques for constructing the face : a polygon mesh and a functional description using a technique called 80ft objects.