Contextual Automatic Speech Recognition on DC
10.94CERMulti-grained
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Multi-grainedTop-K=10, Retrieval=JSPG2026.05 | 10.94 | 11.9 | 10.85 | 86.02 | |
| Multi-grainedTop-K=50, Retrieval=JSPG2026.05 | 11.04 | 12.3 | 10.92 | 85.25 | |
| LLM-RefineTop-K=10, Retrieval=JSPG2026.05 | 11.13 | 12 | 11.05 | 86.17 | |
| Multi-grainedTop-K=100, Retrieval=JSPG2026.05 | 11.22 | 12.9 | 11.07 | 84.8 | |
| LLM-RefineTop-K=50, Retrieval=JSPG2026.05 | 11.32 | 12.7 | 11.19 | 85.09 | |
| LLM-RefineTop-K=100, Retrieval=JSPG2026.05 | 11.36 | 13.1 | 11.2 | 84.98 | |
| Multi-grainedTop-K=1, Retrieval=JSPG2026.05 | 11.4 | 14.76 | 11.09 | 82.05 | |
| LLM-RefineTop-K=1, Retrieval=JSPG2026.05 | 11.51 | 14.5 | 11.24 | 81.56 | |
| CopyNETop-K=10, Retrieval=JSPG2026.05 | 12.1 | 15.55 | 11.79 | 81.33 | |
| CopyNETop-K=50, Retrieval=JSPG2026.05 | 12.18 | 15.9 | 11.84 | 80.8 | |
| CopyNETop-K=1, Retrieval=JSPG2026.05 | 12.25 | 18.9 | 11.65 | 76.62 | |
| CopyNETop-K=100, Retrieval=JSPG2026.05 | 12.26 | 16.1 | 11.91 | 80.44 | |
| Whisper BaseModel=Baseline2026.05 | 12.8 | 22.9 | 11.7 | 71.1 |