摘要 |
PROBLEM TO BE SOLVED: To artificiaily attain an almost natural conversation state with no use of any special device, etc., by changing at least one of shakes of a mouth, head, etc., of a person face image based on the loudness of human voices which are obtained in sequence and at each fixed time interval. SOLUTION: A voice analysis part 744 picks up the changes of voices received from a voice synthesizing part 743 at each prescribed time interval and calculates the mean value of loudness of voices, etc. Then the time series information on each of mean value of voice loudness divided at each time interval of voices is outputted to an image management part 745. The part 745 successively selects the prescribed proper images out of an image data base in response to each information and outputs these images to an image output part 73 to successively display them. These images primarily show an answering person to the voices of a speaker, and the mouths, attitudes, etc., of one or more persons are operated in accordance with the voices produced from a voice output part 72. |