Automatic Speech Recognition on ReazonSpeech (test)
8.4CERNue-ASR
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Nue-ASRModel params.=3,708M, Beam size=1, DeepSpeed-Inference=false2023.12 | 8.4 | 0.22 | |
| Nue-ASRModel params.=3,708M, Beam size=1, DeepSpeed-Inference=true2023.12 | 8.4 | 0.15 | |
| reazonspeech-espnet-v1Model params.=90M, Beam size=202023.12 | 8.7 | 0.27 | |
| reazonspeech-espnet-v1Model params.=90M, Beam size=52023.12 | 9.1 | 0.2 | |
| Whisper-large-v3Model params.=1,541M, Beam size=52023.12 | 11.9 | 0.46 | |
| reazonspeech-espnet-v1Model params.=90M, Beam size=12023.12 | 12 | 0.06 | |
| Whisper-large-v3Model params.=1,541M, Beam size=12023.12 | 12.4 | 0.3 | |
| Whisper-large-v2Model params.=1,541M, Beam size=52023.12 | 20.3 | 0.49 | |
| Whisper-mediumModel params.=769M, Beam size=52023.12 | 20.6 | 0.28 | |
| Whisper-mediumModel params.=769M, Beam size=12023.12 | 21.6 | 0.21 | |
| Whisper-large-v2Model params.=1,541M, Beam size=12023.12 | 21.8 | 0.31 | |
| Whisper-smallModel params.=244M, Beam size=52023.12 | 24 | 0.14 | |
| Whisper-smallModel params.=244M, Beam size=12023.12 | 27.3 | 0.1 | |
| Whisper-baseModel params.=74M, Beam size=52023.12 | 35.4 | 0.08 | |
| Whisper-baseModel params.=74M, Beam size=12023.12 | 38.6 | 0.06 |