Keep what you need : extracting efficient subnetworks from large audio representation models.

David Genova, Philippe Esling, Tom Hurlin

Browse the full ICASSP paper archive.