U.S. Philips Corporation

David A Eves

Richard J Allen

The speech of a group of virtual characters is modelled using a directed graph or audio web, where each utterance is a node (A1-A7, B1-B5) in the web having links from it to each possible utterance that can be made in reply at any point in the course of the interaction. The choice of the next node (conversation segment) is dependent on attributes of the character and/or the virtual environment in which the conversation is taking place. These attributes may be fixed, modified by system function calls available at each node, or modified in response to user input.

Growing community of inventors

Crawley, United Kingdom

David A Eves