Code-switching Automatic Speech Recognition on ViMedCSS Vietnamese-English medical domain (test)
21.87WERWCE + CL (tri-level)
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| WCE + CL (tri-level)Backbone=Whisper-small, Adaptation=LoRA, Filtering Strategy=tri-level (Acoustic + Text + Phoneme), Loss=WCE + Contrastive Alignment2026.06 | 21.87 | 18.74 | |
| WCE + CL (N-best NM)Backbone=Whisper-small, Adaptation=LoRA, Filtering Strategy=N-best NM, Loss=WCE + Contrastive Alignment2026.06 | 22.86 | 19.1 | |
| CE + CL (N-best NM)Backbone=Whisper-small, Adaptation=LoRA, Filtering Strategy=N-best NM, Loss=CE + Contrastive Alignment2026.06 | 23.16 | 20.11 | |
| MWERBackbone=Whisper-small, Adaptation=LoRA, Loss=Sequence-level baseline2026.06 | 23.82 | 20.84 | |
| WCEBackbone=Whisper-small, Adaptation=LoRA, Loss=Weighted cross-entropy2026.06 | 24.21 | 21.18 | |
| CEBackbone=Whisper-small, Adaptation=LoRA, Loss=Standard cross-entropy2026.06 | 24.72 | 21.95 |