Text-to-gesture system for embodied agents

This paper presents a Text-To-Gesture (TTG) system which enables an embodied agent to generate proper gestures for a given text input. To transform text information to a sequence of gestural motions, it is important to analyze coverbal gestures linguistically. It should be also concerned to reproduce natural motions synchronized with speech signals in a human-like fashion. This paper proposes a TTG system which consists of morphological decomposition, gesture selection, and gesture scheduling module. The proposed system is tested through a storytelling example for an embodied agent.