Marvin Lavechin

Meta AI, Ecole Normale Supérieure

picture.jpg

Centre Sciences des Données

45 rue d'Ulm

75005, Paris, France

I am a PhD student under the supervision of Alejandrina Cristia, Hervé Bredin and Emmanuel Dupoux. I work between Meta AI and the Cognitive Machine Learning and Language Acquisition Across Cultures teams of the LSCP.

My work is at the intersection of Artificial Intelligence and Cognitive Sciences. My research focuses on language acquisition modelling and speech processing applications to the study of language development in general.

Broad research questions include: How can we build machines that learn to perceive language? How can we better compare humans and machines in language-perceptual tasks? What makes infants so efficient at learning language?

News

Nov, 2023 [🇫🇷 SEMINAR] I’m participating to the GDR LIFT TAL seminar. See you in Marseille!
Oct, 2023 [📝 SUBMISSION] Our work Brouhaha has been submitted to ICASSP 2023 !
Sep, 2023 [🇩🇪 SEMINAR] I’m participating to the Developmental Machine Learning: From Human Learning to Machines and Back seminar. See you in Dagstuhl!
Aug, 2023 [🇫🇮 VISIT] Two-week visit in Okko Räsänen’s lab, Tampere, Finland!
Mar, 2023 [🤖 SEMINAR] Kick-off meeting of the Benchmarking speech acquisition in humans & machines online seminar (click here to join us!)

Selected publications

  1. Brouhaha: multi-task training for voice activity detection, speech-to-noise ratio, and C50 room acoustics estimation
    Marvin Lavechin, Marianne Métais, Hadrien Titeux, and 7 more authors
    Submission to ICASSP 2023
  2. Statistical learning models of early phonetic acquisition struggle with child-centered audio data
    Marvin Lavechin, Maureen Seyssel, Marianne Métais, and 5 more authors
    Submission to Cognition 2022
  3. Reverse engineering language acquisition with child-centered long-form recordings
    Marvin Lavechin, Maureen Seyssel, Lucas Gautheron, and 2 more authors
    Annual Review of Linguistics 2022
  4. An open-source voice type classifier for child-centered daylong recordings
    Marvin Lavechin, Ruben Bousbib, Hervé Bredin, and 2 more authors
    Interspeech 2020
  5. End-to-end domain-adversarial voice activity detection
    Marvin Lavechin, Marie-Philippe Gill, Ruben Bousbib, and 2 more authors
    Interspeech 2020