Visual text reader for virtual image communication on networks

This paper presents a media conversion system from text to video, which can be used as a virtual image communication tool over narrow-band networks. The proposed system analyzes a plain text, such as e-mail, and generates a video sequence for a human's bust shot which includes actions and facial expressions related to the contents. The voice sounds are also generated using text-to-speech system. Both video and voice are synchronized, so that video-phone like communication is available without any additional information out of the original text.