Post-ASR Entity Correction on Common Voice
14.45WERLLM-Select
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| LLM-SelectLLM=GPT-4o, Correction Strategy=LLM-Select2026.03 | 14.45 | 13.88 | 45.7 | 84.13 | 87.74 | 80.79 | |
| 1-BestLLM=GPT-4o, Correction Strategy=1-Best2026.03 | 14.82 | 15.05 | 41.2 | 81.06 | 82.63 | 79.57 | |
| BaselineLLM=N/A, Correction Strategy=Baseline2026.03 | 15.02 | 25.57 | — | 73.4 | 96.39 | 59.26 | |
| LLM-SelectLLM=GPT-4o-mini, Correction Strategy=LLM-Select2026.03 | 15.88 | 21.56 | 15.7 | 76.96 | 91.48 | 66.43 | |
| ROVER EnsembleLLM=GPT-4o, Correction Strategy=ROVER Ensemble2026.03 | 18.17 | 15.69 | 38.6 | 80.09 | 81.66 | 78.58 | |
| Entity-Aware SelectLLM=GPT-4o, Correction Strategy=Entity-Aware Select2026.03 | 18.72 | 15.35 | 40 | 79.98 | 80.43 | 79.53 |