Post-ASR Entity Correction on ContextASR-Bench
9.3WERLLM-Select
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| LLM-SelectLLM=GPT-4o, Correction Strategy=LLM-Select2026.03 | 9.3 | 3.97 | 17.8 | 95.17 | 94.95 | 95.4 | |
| BaselineLLM=N/A, Correction Strategy=Baseline2026.03 | 9.4 | 4.82 | — | 94.89 | 95.26 | 94.51 | |
| 1-BestLLM=GPT-4o, Correction Strategy=1-Best2026.03 | 9.42 | 4.21 | 12.7 | 94.82 | 94.49 | 95.14 | |
| LLM-SelectLLM=GPT-4o-mini, Correction Strategy=LLM-Select2026.03 | 9.75 | 4.51 | 6.6 | 94.65 | 94.44 | 94.86 | |
| ROVER EnsembleLLM=GPT-4o, Correction Strategy=ROVER Ensemble2026.03 | 9.77 | 3.82 | 20.8 | 94.81 | 94.08 | 95.54 | |
| Entity-Aware SelectLLM=GPT-4o, Correction Strategy=Entity-Aware Select2026.03 | 10.54 | 4.36 | 9.6 | 94.33 | 93.78 | 94.89 |