arXiv - CSCL: "4D ASR: Joint modeling of CTC, Attention, Transdu…" - Qoto Mastodon

arXiv - CSCL @arxiv_cscl@qoto.org

4D ASR: Joint modeling of CTC, Attention, Transducer, and Mask-Predict decoders. (arXiv:2212.10818v2 [cs.SD] UPDATED)

http://arxiv.org/abs/2212.10818 #arXiv #NLProc

May 31, 2023, 03:10 · · arxiv-cscl · · ·

Sign in to participate in the conversation