Automatic Speech Recognition on UIT-ViMD (test)
12.58WEROur-Conformer
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| Our-ConformerSetup=Fully Supervised, Params=28M2026.05 | 12.58 | — | 10.88 | 14.69 | 13.58 | 0.28 | |
| Our-TransformerSetup=Fully Supervised, Params=26M2026.05 | 12.59 | — | 10.77 | 14.8 | 13.68 | 0.33 | |
| Our-TransformerSetup=P+V, Params=26M, Time(s)=0.332026.05 | 13.05 | 8.36 | 11.35 | 15.37 | 14.05 | — | |
| Our-ConformerSetup=P+V, Params=28M, Time(s)=0.282026.05 | 13.27 | 8.47 | 11.29 | 16.21 | 14.08 | — | |
| wav2vec2-base-vi-250hSetup=FT, Params=95M2026.05 | 13.56 | — | 12.29 | 17.15 | 15.26 | — | |
| wav2vec2-base-vi-250hSetup=Fine-Tuning, Params=95M2026.05 | 13.56 | — | — | — | — | — | |
| ConformerSetup=P, Params=28M, Time(s)=0.932026.05 | 13.89 | 11.22 | 12.51 | 17.52 | 15.22 | — | |
| TransformerSetup=P, Params=26M, Time(s)=1.252026.05 | 15.53 | 12.47 | 13.31 | 15.77 | 16.46 | — | |
| wav2vec2-base-viSetup=FT, Params=95M2026.05 | 15.8 | — | 14.64 | 20.27 | 17.72 | — | |
| wav2vec2-base-viSetup=Fine-Tuning, Params=95M2026.05 | 15.8 | — | — | — | — | — | |
| MT-TransformerSetup=P+M, Params=26M, Time(s)=1.582026.05 | 15.9 | 12.95 | 14.27 | 17.56 | 17.15 | — | |
| PhoWhisper-baseSetup=FT, Params=74M2026.05 | 16.3 | — | 13.2 | 18.26 | 13.54 | — | |
| PhoWhisper-baseSetup=Fine-Tuning, Params=74M2026.05 | 16.3 | — | — | — | — | — | |
| TransformerSetup=Fully Supervised, Params=29M2026.05 | 16.66 | — | 14.68 | 18.64 | 17.95 | 0.37 | |
| TransformerSetup=Scratch, Params=29M, Time(s)=0.372026.05 | 16.71 | — | 14.32 | 18.24 | 17.49 | — | |
| MT-ConformerSetup=P+M, Params=28M, Time(s)=1.212026.05 | 16.77 | 13.5 | 14.63 | 19.53 | 17.71 | — | |
| wav2vec2-base-vi-250hSetup=Zero-shot, Params=95M2026.05 | 16.83 | — | 14.51 | 19.73 | 18.23 | 0.07 | |
| PhoWhisper-baseSetup=Zero-shot, Params=244M2026.05 | 17.11 | — | 15.25 | 18.5 | 18.38 | 1.21 | |
| wav2vec2-base-vi-250hSetup=Zero-shot, Params=95M, Time(s)=0.072026.05 | 17.47 | — | 14.98 | 20.97 | 17.24 | — | |
| wav2vec2-base-vi-160hSetup=FT, Params=95M2026.05 | 17.49 | — | 16.7 | 24.56 | 20.51 | — | |
| ConformerSetup=Scratch, Params=31M, Time(s)=0.292026.05 | 17.49 | — | 16.71 | 19.68 | 18.52 | — | |
| wav2vec2-base-vi-160hSetup=Fine-Tuning, Params=95M2026.05 | 17.49 | — | — | — | — | — | |
| ConformerSetup=Fully Supervised, Params=31M2026.05 | 17.56 | — | 16.05 | 19.92 | 18.39 | 0.29 | |
| PhoWhisper-baseSetup=Zero-shot, Params=74M, Time(s)=0.732026.05 | 18.61 | — | 14.69 | 24.15 | 17.87 | — | |
| Whisper-baseSetup=FT, Params=74M2026.05 | 19.93 | — | 20.05 | 27.89 | 20.89 | — | |
| PhoWhisper-smallSetup=Zero-shot, Params=74M2026.05 | 21.78 | — | 18.17 | 27.18 | 23.8 | 0.73 | |
| wav2vec2-base-viSetup=Zero-shot, Params=95M2026.05 | 22.48 | — | 19.87 | 27.15 | 23.82 | 0.03 | |
| wav2vec2-base-viSetup=Zero-shot, Params=95M, Time(s)=0.032026.05 | 23.07 | — | 20.32 | 27.28 | 22.48 | — | |
| wav2vec2-base-vi-160hSetup=Zero-shot, Params=95M2026.05 | 31.12 | — | 27.09 | 40.49 | 32.83 | 0.03 | |
| Whisper-baseSetup=Zero-shot, Params=244M, Time(s)=1.212026.05 | 31.38 | — | 26.37 | 39.91 | 29.46 | — | |
| wav2vec2-base-vi-160hSetup=Zero-shot, Params=95M, Time(s)=0.032026.05 | 31.74 | — | 27.5 | 38.12 | 30.93 | — |