抄録
People can enjoy Japanese traditional comic storytelling by both hearing the voice and watching the gesture of a teller. Some records of comic storytelling are available only by a voice record. If the animated motion for the talk is generated, we may enjoy storytelling more. We propose two methods to generate animation in accordance with a talk. One is a real time teaching; animation is generated at real time using a mouse. The other is a semiautomatic method; the timing for a new motion will be searched and the candidates of possible motion patterns are shown to an operator. The operator will choose a motion pattern fitted with the talk of a teller. Experimental results showed the effectiveness of these methods. The methods will be applicable to the development of animation fitted with voice or music data.
