Virtual Talk: a model-based virtual phone using a layered audio-visual integration

An implementation of a one-to-one model based virtual phone system, the Virtual Talk, is presented. In addition to the conventional ability of Internet phone applications to transmit speech messages between two users, the facial expressions of each user is real-time analyzed, transmitted, and synthesized in a model based approach. Thereby, a joint audiovisual communication can be achieved in less than 10 kbps. To resolve the problem of audio-visual synchronization, a layered architecture is also proposed for audio-visual integration