Recognition and interpretation on talking agents
Speech is not a common way to interact with an artwork, but recent advances in Artificial Intelligence imply challenges and opportunities to explore new kinds of experiences in the confrontation of the spectator to the artwork. In order to have flexibility in experimentation by performing different art-installations, we have developed and implemented the concept of Talking Agent, a reusable software component with the ability to recognize human speech and synthesize a spoken response, and to make use of other sensors and actuators. The paper describes how the speech recognition and interpretation process of the Talking Agents have been studied in order to develop alternatives to obtain acceptable results in the desired scenarios.