UMA-Split: unimodal aggregation for both English and Mandarin non-autoregressive speech recognition

Open in new window