Efficient Dialogue Management Of A Spoken Dialogue System By Using Prosody Information Of The User Utterances

Words used by a human while framing a response during the interaction with a software agent like spoken dialogue system(SDS) has valuable information as regards internal mental state of the user is concerned. The users level of certainty about a response could be judged by the prosody information structure. Prosody reveals Information about the context by highlighting information structure and aspects of the speaker hearer relationship. Most often it is observed that the speakers internal state is not depicted by the words he uses but by the tone of his utterance or facial expression of the user. This paper uses PRATT a tool for speech analysis which uses 15 acoustic features to determine the certainty of the responses of the user using the prosody information which will actually aid the dialogue management component of the SDS in framing a better dialogue strategy.