A Physics-based Talking Head for Interface Agent

Computer graphics is vital for the depiction of life-like agents for computer interfaces and visual entertainment systems. This paper develops a realistic graphical model of the human face capable of synthesizing facial expressions in natural conversation scenarios. In particular, we propose a facial muscle model that includes oral musculature for controlling mouth shapes during speech production. Basic mouth shapes are defined through the measurement of natural human facial photographs captured by two cameras. We animate speech by realizing lip synchronization using standard phoneme durations.