Hierarchy and fluctuation in time structure of spoken dialogue

We assume that natural spoken dialogue cannot be realized without non-verbal information and simultaneous interaction. Therefore, in order to make a spoken dialogue system having these properties, we consider the rhythm in non-verbal information of speech. This is because the system can simultaneously interact with man by using entrainment of such rhythm. In this paper, we create some speech models by rhythm and construct simple spoken dialogue system. As a result of the experiments using this system, some of the time structures of spoken dialogue are clarified. It is assumed that spoken dialogue has hierarchy in time structure and fluctuation emerges under simultaneous interaction.