Second ESCA/IEEE Workshop on Speech Synthesis
September 12-15, 1994
This paper presents an implemented system for automatically producing prosodically appropriate speech and corresponding facial expressions for animated, three-dimensional agents that respond to simple database queries. Unlike previous text-to-facial animation approaches, the system described here produces synthesized speech and facial animations entirely from scratch, starting with semantic representations of the message to be conveyed, which are based in turn on a discourse model and a small database of facts about the modeled world.
Bibliographic reference. Pelachaud, Catherine / Prevost, Scott (1994): "Sight and sound: generating facial expressions and spoken intonation from context", In SSW2-1994, 216-219.