I am in the process of completing a proof-of-concept for a voice-enabled rule-based system, i.e., it uses voice recognition (with a command & control grammar) to drive the interaction with a rules-based intelligent agent and voice synthesis for the agent to respond verbally with the user.
 
So far, the results have been encouraging and demonstrate the feasibility of man-to-machine communication using common-off-the-shelf (COTS) software. The next step is to integrate a FIPA Agent Communication Language (ACL) to express a knowledge representation to the agent so that it can learn (from the user) by "hearing" what the user knows.
 
Has anyone out there been doing anything like this ?

Reply via email to