Section: Contracts and Grants with Industry

Regional Actions

CPER MISN TALC

The team is involved in the management of the Contrat Plan Etat-Région (CPER) contract. In particular, Christophe Cerisara is co-responsible, with Claire Gardent, of the CPER MISN TALC, which objective is to leverage collaborations between regional academic and private partners in the domain of Natural Language Processing and Knowledge engineering. The TALC action involves about 12 research teams and 40 researchers for a budget of about 240,000 euros per year.

In addition to the co-management of this project, our team is also involved in two scientific collaborative operations:

• An operation about text-to-speech alignement, in collaboration with the TALARIS research team and the ATILF laboratory. This operation aims at proposing semi-supervised solutions to facilitate the transcription and processing of large bimodal text and speech corpora. The main outcomes of this operation are (1) the JTrans software described in section  5.6 , and a concordancer that was developped in Java by two BSc students in the framework of their final year project.

• An operation about syntactic analysis of speech transcripts, in collaboration with the TALARIS research team and the ATILF laboratory. This operation aims at adapting state-of-the-art stochastic parsers to the specificities of manual and automatic transcriptions of speech, and at building a French treebank of broadcast news speech transcripts. The main outcome of this operation is the J-Safran software, described in section  5.5 .

“Intonale”: Perception and production of prosodic contours in L1 and L2

This action, launched by the CCOSL, aims at developping collaboration between academic partners from Lorraine laboratories and universities. It has started in september 2009 and should last until the end of 2011. The speech team from LORIA is associated with the laboratory ATILF (Mathilde Dargnat). The project deals with the perception and production of prosodic contours in the first language (L1) and in a second language (L2). We have chosen two radically different languages with respect to prosody : French and English. We have collected a corpus recorded by 34 French speakers and made up of sentences with different modalities: assertions, questions, major and minor continuations. French speakers uttered these sentences both in French (their native language) and in English (the “targeted” non native language). The English part of the corpus is used by the project ALLEGRO, presented hereafter. The French part of the corpus is currently segmented, whilst its English part is segmented under the framework of the INTERREG project ALLEGRO. In order to record corpora in other languages, we improved the Corpus Recorder software (see 5.9 ). The previous corpus had also been recorded by native english speakers (French and English sentences).