Labiophone
The Labiophone was initiated by Christian
Benoit at ICP and became rapidly a major project in the ELESA
federation. The general aim of the project is to capture and characterize
lip motion of a subject and use these characteristic parameters in speech
technology. Applications are numerous:
-
Audio-visual Speech Coding: lip parameters may be used to animate through
the telecommunication network a virtual clone of the subject. Animation
of talking faces is typically part of the MPEG4 project.
-
Robust Audio-visual Speech Recognition: lip movements may enhance the robustness
of current speech recognizers by providing both additional voice localization/separation
and articulation cues.
References
-
Gérard Bailly. Le labiophone Document de Synthèse,
Projet soumis à la région Rhones-Alpes, Institut de la Communication
Parlée, Grenoble, France, 1998. (Word,
7 pages, 8095232 bytes)
-
Gérard Bailly. Le labiophone et MPEG4, Présentation à
l'occasion de la visite de Weinfeld & Saglio (
intro_spi07_00.PDF,
spi07_00.PDF,
spi.zip)
Demos
-
Tracking facial movements with an analysis-by-synthesis technique
original.avi
, original sequence
wire.avi,
wireframe of the 3D model of facial articulation (produced by MOTHER)
superposed
bise.avi
, wireframe of the 3D model of facial articulation for a whole paragraph
reading
Retour à la page de G. Bailly
Retour à l'index de l'ICP
Last
modified: Thu Feb 11 13:44:53 MET 1999