LeBenchmark 2.0: A standardized, replicable and enhanced framework for self-supervised representations of French speech

Titouan Parcollet, Ha Nguyen, Solène Evain, Marcely Zanon Boito, Adrien Pupier, Salima Mdhaffar, Hang Le, Sina Alisamir, Natalia A. Tomashenko, Marco Dinarelli, Shucong Zhang, Alexandre Allauzen, Maximin Coavoux, Yannick Estève, Mickael Rouvier, Jérôme Goulian, Benjamin Lecouteux, François Portet, Solange Rossato, Fabien Ringeval et al. (2 additional authors not shown)

Abstract: Highlights•Open-source framework for assessing self-supervised representations in the French language.•14,000 h of heterogeneous speech documented into four datasets.•14 pre-trained self-supervised models for French, ranging from 26 to 965 million neural parameters.•6 standardized tasks for the evaluation of French self-supervised models.