DEEP VOICE 2022

  • Symposium

Deep Voice, Paris is the Parisian event dedicated to the voice and artificial intelligence with the aim of bringing together scientific and technical expertise, innovation, and entrepreneurship. For this second edition, keynotes, discussions, and panel discussions will address diversity and inclusion in voice technologies: rare languages, low-resource languages, sign language, dialects, bias, and discrimination. 

 Technical hands-on workshops will be offered to participants to further enhance their skills in speech processing and natural language.

At the end of the day, a networking session will be held so participants can meet and exchange ideas in a relaxed atmosphere.

Deep Voice is an event co-organized by IRCAM, Sorbonne University, and SCAI as part of the ManiFeste-2022 festival.
For this second edition: even more discussions, more meetings, more events with the best French and international specialists in the field, technical training workshops and cultural and artistic openings!

Organizing Committee:
Nicolas Obin, STMS Lab (IRCAM, CNRS, Sorbonne University)
Xavier Fresquet, SCAI (Sorbonne Center for Artificial Intelligence, Sorbonne University)

PROGRAM

Wednesday, June 15 (2:00-5:00 pm, location: SCAI)

- 2:00-2:15 PM, Introduction, Nicolas Obin, STMS (IRCAM, Sorbonne University, CNRS, French Ministry of Culture) and Xavier Fresquet, Sorbonne Center for Artificial Intelligence, Sorbonne University

- 2:15-3:30 pm, Keynote "History of voice technologies, current developments and perspectives
Speaker: Luc Julia

Moderator: Nicolas OBIN, STMS (IRCAM, Sorbonne University, CNRS, French Ministry of Culture)

- 3:30-5:00 pm, Discussion (in English): "What about my voice, what does it have as a gender?"
Abstract: What are the relationships between voice and gender? How does gender interact with identity and emotion? How do we consider gender in voice technologies?

Speakers: Emil Asmussen, Creative Director of Meet Q, The World's First Genderless Voice, Melissa A. Barkat-Defradas, Institute of Evolutionary Science, Nadine Lavan, Queen Mary University of London, Giulia Perugia, Eindhoven University of Technology (TU/e) - Eindhoven (Netherlands)

Moderators: Victor Rosi (IRCAM), Sylvie Saget (University of Gothenburg)

- 5:00-6:30 pm: Reception at the SCAI, evening meetings and networking

Thursday June 16 (10am-1pm location: SCAI, then 2pm-5pm, location: IRCAM)

- 10am-1pm at SCAI
Workshop #1: Hands-on speech-to-text with neural networks

Speakers: Yann Teytaut and Clément Le Moine Veillon, STMS (IRCAM, Sorbonne University, CNRS, French Ministry of Culture)

 

- 2:00-3:30 pm at IRCAM
Discussion: Diversity of languages in the digital world
Abstract: What representation of minority languages in speech technologies? How can technology help preserve languages and cultures?

Speakers: Laurent Besacier, GIPSA lab, Victoria Khurshudyan, INALCO
Moderators: to be announced

 

- 3:30-6pm at IRCAM
Sign language session + SLAM performance

Participants: Agnès Vourc'h, CNRHR Robert Laplane, Michael Filhol, LIMSI, Djenebou Bathily, Université Vincennes - Saint-Denis

Evening 2: to be announced

 

Friday June 17 (10am-1pm location: SCAI, then 2pm-5pm, location: IRCAM)

  - 10am-1pm at SCAI

Workshop #2: Hands-on CamemBERT

Speakers: to be announced

 

- 2pm-3:30pm at IRCAM
Discussion: Is bias in the text?

Speakers: Eric de la Clergerie, Inria, Djamé Seddah, Sorbonne University
Moderator: Laure Soulier, Sorbonne University

 

- 3:30-5:00 pm at IRCAM
Panel discussion (in English): Dialectal and accentual diversity in speech technologies

Speakers: Mathieu Avanzi, Sorbonne University, Anton Lozhkov, Hugging Face, Maxim Serebryakov, SANAS
Moderator: Carl Robinson (Rumble studio)

 

- 5-5:15 pm at IRCAM
Concluding remarks, Nicolas Obin, STMS (IRCAM, Sorbonne University, CNRS, French Ministry of Culture) and Xavier Fresquet, Sorbonne University, SCAI

 

https://www.sorbonne-universite.fr/evenements/deep-voice-paris

 

En poursuivant votre navigation sur ce site, vous acceptez l'utilisation de cookies pour nous permettre de mesurer l'audience, et pour vous permettre de partager du contenu via les boutons de partage de réseaux sociaux. En savoir plus.