DeCRED: Decoder-Centric Regularization for Encoder-Decoder Based Speech Recognition Paper • 2508.08938 • Published Aug 12 • 11
DeCRED: Decoder-Centric Regularization for Encoder-Decoder Based Speech Recognition Paper • 2508.08938 • Published Aug 12 • 11 • 2
DiCoW Collection DiCoW (Diarization-Conditioned Whisper) is a collection of speaker-aware ASR models developed by BUT-FIT, extending OpenAI’s Whisper. • 5 items • Updated Jul 15 • 1