Year: 2026
Type: preprint
Abstract: Speech self-supervised learning aims to learn general-purpose representations for downstream speech tasks. However, current approaches rely on complex, carefully designed prediction targets. We challenge this necessity with GLaS-JEPA, a framework that directly predicts the current encoder's continuous representations at masked positions, without contrastive learning, discrete targets, or separate EMA target encoders. We prevent representation collapse using SIGReg representation-space regularization, eliminating the need for engineered target-generation mechanisms. Pretrained on 960 hours of LibriSpeech, our 57M-parameter model achieves a 6.89% WER on frozen-encoder SUPERB ASR and a 25.87% CER on slot filling, outperforming the best non-distilled sub-90M baselines by 43.1% and 22.0%, respectively. These results demonstrate that highly competitive speech representations can emerge from a radically simplified training recipe.
Source: arXiv (Cornell University)
Institutions: Mila - Quebec Artificial Intelligence Institute, Concordia University, Centre National de la Recherche Scientifique, Université de Toulon, Aix-Marseille Université, Laboratoire d’Informatique et Systèmes, Centre Inria de l'Université Grenoble Alpes, Laboratoire Jean Kuntzmann, Université Grenoble Alpes, Université Laval, Institut polytechnique de Grenoble, GIPSA-Lab
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7214968424