A Quantitative Analyzing Method for the Synchrony Between a Speaker's Voice and a Listener's Nodding in Interpersonal Communication

抄録

A quantitative analyzing method for the synchrony between a speaker's voice and a listener's nodding in interpersonal communication is proposed. A parameter of nodding is defined as the total area of a subtracted image between successive binary images converted from video frames. A parameter of voice is defined as the difference between the maximum and the minimum amplitude of the voice signal in the interval corresponding to the frame image. The relation between these two parameters is evaluated by the cross-correlation function. By using this method, basic characteristics of this synchrony are discussed. In addition, a real-time analyzing system for this voice-movement synchrony is developed for the purpose of processing a large amount of data. This study is useful for many applications, especially to realize a man-machine interface with smooth information exchange.