 |
ESCA Workshop on Spoken Dialogue SystemsVigsø, Denmark |
 |
|
ESCA Workshop on Spoken Dialogue SystemsVigsø, Denmark |
|
Communication between humans do not only make use of voice but is normally based on several other perception and production capabilities (gesture, vision, touch, etc.). In order to render human interaction with the machine more efficient and close to the human model, the tendency is now to integrate several different means of communication. But if humans skillfully combine gestures, voice production and hearing, when communicating with each other, the machine does not possess the same type of competence. Several problems occur at a low level due to device response times and time management. At a higher level, one major requirement is therefore to accurately model knowledge at all levels. The present study proposes an architecture which attempts to take some of those problems into account, attributing specific roles to the different controllers which have been identified.
Bibliographic reference. Bellik, Y. / Ferrari, S. / Néel, Françoise / Teil, D. (1995): "Requirements for multimodal dialogue including vocal interaction", In SDS-1995, 161-164.