Abstract: A computer based system utilizing cues extracted from audio speech to select from among a database of stored image cells to produce synthesized animated characters. A database of image cells is created by storing the face of an actor or other character articulating some 40 phoneme sounds, in 8 different emotions from 25 different camera angles. Audio speech is then analyzed and the faces are extracted based on phoneme sounds and emotion cues within the spectral information of the speech. The position of a user operated camera controller is polled to determine a desired camera angle and the three image parameters (articulation, emotion and camera angle) are used to select an image from the database associated with the phoneme. The selected images are interpolated to produce a smooth, life-like character or cartoon animation sequence, and then stored to a video recorder.