Sub-project 02

Objective: collect and produce corpora that will be used for learning, producing and assessing the automatic generation of captions, subtitles and French Sign Language (LSF).

Under the responsibility of france.tv access, this module consists in collecting the language resources used in sub-projects 3, 4 and 5.

These resources consist of already existing corpora, which can be used in the context of the project. These corpora include subtitled television programs or French-language dubbing, and also include schemes created by deaf translators for the transition from written French into French Sign (LSF) language generation, and research corpora.

MOCAPLAB will complete these resources by recording a 3D corpora of LSF movements. These movements are necessary for the animation of virtual signers, as well as a corpora of translation diagrams produced by the deaf translators used during the project.