We present an automatic technique for 3D surface reconstruction of a human face and a suitable semiautomatic adaptation method for a real time animation of the audiovisual speech synthesis. The face capture employs photogram-metric methods on the basis of calibrate stereoscopy and a projected structured light. The method generates successive range data. A generic face model is used and it is modified to fit to a 3D measurement data of the specific face. The teeth, tongue and eyes are automatically added to the model and all textures are mapped. This individual model is used in a real time animation system. The animation is applied in a visual part of a speech dialog.
Cite as: Krnoul, Z., Zelezný, M., Cisar, P. (2004) Face model reconstruction for Czech audio-visual speech synthesis. Proc. 9th Conference on Speech and Computer (SPECOM 2004), 47-51
@inproceedings{krnoul04_specom,
author={Zdenek Krnoul and Milos Zelezný and Petr Cisar},
title={{Face model reconstruction for Czech audio-visual speech synthesis}},
year=2004,
booktitle={Proc. 9th Conference on Speech and Computer (SPECOM 2004)},
pages={47--51}
}