Automatic Speech Recognition on L2-ARCTIC (Unseen-transcript (UT))
9.14WERW2V2-Large + SupCon
Evaluation results
| Method | Links | |
|---|---|---|
| W2V2-Large + SupConBackbone=wav2vec 2.0 Large, Regularization=Supervised Contrastive (SupCon)2026.05 | 9.14 | |
| W2V2-Large (CTC)Backbone=wav2vec 2.0 Large, Objective=CTC2026.05 | 10.47 | |
| MAS-LoRA-QKVOModel Type=MAS-LoRA-QKVO2026.05 | 11.77 | |
| Whisper FTModel Type=Whisper Fine-tuned2026.05 | 12.21 |