Different experienced and internationally recognized researchers expressed their willingness to participate in the ANR DIETS project. A presentation of the international team as well as the implication in the project of each researcher is summarized here:
Richard Dufour is the coordinator of the DIETS project. He is Associate Professor at the Laboratoire Informatique d’Avignon (LIA – Avignon University) since 2012. He obtained his Ph.D. on Automatic transcription of spontaneous speech from Le Mans University on the 1st December 2010. He worked 4 months as a research intern at M*Modal (Pittsburgh – USA), a company specialized in medical transcriptions. He spent one year as a postdoctoral researcher in the language processing team at Orange Labs (Lannion – France). His main research interests concern automatic speech processing, with a major part of works on automatic transcription quality.
Jane Wottawa is an associate Professor in French Linguistics at the LIUM (Le Mans University) since 2018. The researcher defended her PhD thesis on speech production and perception in French learners of German in 2017. During her post-doc (2018) at LIMSI – CNRS she worked on ASR transcription errors in French and the automatic transcription of non-native pronunciation variants in French and Arabic. She will bring expertise on transcription errors and on their linguistic categorization.
Arnaud Rey is a senior CNRS researcher at the Cognitive Psychology Laboratory (UMR 7290, Marseille). His work focuses on the perception and production of language in an approach that combines experimentation and modeling. Junior member of the Institut Universitaire de France in 2007, he received an early academic award from the Association for Psychological Science in 2009. His most recent publications appeared in Consciousness and Cognition (2019), Topics in Cognitive Science (2018), Cognitive Science (2015, 2018), PLoS ONE (2017), NeuroImage (2016). He will be a mentor in the project on cognitive aspects.
Yannick Estève is a Professor in Computer Science at the LIA (Avignon University). His current research activities focus on spoken language understanding, speech analytics, speech recognition, neural networks, and word embeddings. He is currently head of the LIA. He is leading the ON-TRAC project (2019-2022) funded by the ANR on end-to-end speech translation, and he is the LIA scientific responsible for the ANR DEEP-PRIVACY project focusing on privacy in the framework of speech recognition. He will be a mentor on end-to-end approaches.
Thierry Bazillon is a data scientist expert at Allo-Media, a company specialized in automatic speech conversation analysis. He holds a Ph.D. in Language Science and worked on transcribing and manually processing spontaneous speech for its automatic recognition. He worked on the development of several corpora intended for scientific research, such as the ANR EPAC and DECODA corpus. Thierry is at the initiative of the AlloSat corpus, freely distributed, which the DIETS project will use as a primary resource. He will give support to the corpus, and his expertise on annotation and collection.
The involvement of this team of researchers with complementary skills is an undeniable asset of the project and the work that is carried out there. Indeed, this innovative project is at the border of several research disciplines, ranging from linguistics to cognitive sciences, via machine learning, and it seems important, even essential to its success, to be able to benefit from the skills recognized by each expert in his area of competence.