Long-form Transcription on CORAAL
16.2WERWhisper
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| WhisperBackbone=Large V2, Evaluation Protocol=Zero-shot2022.12 | 16.2 | 54.5 | |
| Whisper large-v22022.12 | 19.6 | — | |
| Whisper medium2022.12 | 20.3 | — | |
| Whisper large2022.12 | 20.4 | — | |
| Whisper medium.en2022.12 | 20.6 | — | |
| Whisper small.en2022.12 | 21.9 | — | |
| Whisper small2022.12 | 22.3 | — | |
| stt_en_conformer_ctc_large2022.12 | 25.1 | — | |
| Whisper base.en2022.12 | 25.2 | — | |
| Whisper base2022.12 | 26.9 | — | |
| Whisper tiny.en2022.12 | 30.3 | — | |
| Whisper tiny2022.12 | 33.1 | — | |
| wav2vec 2.0Backbone=Large, Language Model=None2022.12 | 35.6 | — | |
| wav2vec2-large-960h-lv60-self2022.12 | 36.7 | — | |
| wav2vec2-large-robust-ft-libri-960h2022.12 | 36.8 | — | |
| hubert-large-ls960-ft2022.12 | 37.9 | — | |
| hubert-xlarge-ls960-ft2022.12 | 38.1 | — | |
| stt_en_conformer_transducer_xlarge2022.12 | 38.9 | — | |
| wav2vec2-large-960h2022.12 | 43.5 | — | |
| wav2vec2-base-960h2022.12 | 49.1 | — | |
| wav2vec2-base-100h2022.12 | 55.4 | — |