A conversation robot with back-channel feedback function based on linguistic and nonlinguistic information

A conversation robot which can generate a back-channel feedback appropriately to the user is developed. The contents of these feedbacks depend on the contents of the dialogue partner’s utterance, which are provided by the speech recognition. In order to determine the content of the feedback earlier than the end of the utterance, we used finite state transducer based speech recognizer which outputs the content of the feedback. And we used prosody information, especially the fundamental frequency(F0) and the power of the utterance, to extract the proper timing of the feedback. We implemented these modules and applied them to the spoken dialogue system on the humanoid robot ROBISUKE. Experimental results show the effectiveness of our methods