擬人化エージェントにおける自然さを考慮した口唇画像の簡易発話モデル

概要

論文の詳細を見る
The multimodal anthropomorphic agent interfaces are effective means as a human communication technology. This paper has proposed the simple utterance model of the mouth shape for applying a multimodal anthropomorphic agent to small-scale system environment. We consider that the sequence of the height/width of a mouth shape measured from a Japanese newscaster's mouth movement image video is a signal waveform. The frequency ingredients contained in these waveforms are analyzed by Fourier transform. We have created the simple utterance model that is the typical waveform of the height/width of a mouth shape calculated from only nine or ten frequency ingredients. The validity of our proposed utterance model is verified by the comparison examination for two conventional simple utterance animation models.
日本感性工学会の論文