Automatic Speech Recognition on M3AV (test)
9.3WER (All)2-stage Reason (ours)
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| 2-stage Reason (ours)Train set=M, Prompting method=2-stage Reason (ours)2026.06 | 9.3 | 23.1 | 23.3 | |
| 2-stage TranscribeTrain set=M, Prompting method=2-stage Transcribe2026.06 | 9.4 | 24 | 23.9 | |
| 2-stage Reason (ours)Train set=L, Prompting method=2-stage Reason (ours)2026.06 | 9.5 | 23.4 | 23.6 | |
| Transcribe w/ ContextTrain set=M, Prompting method=Transcribe w/ Context2026.06 | 9.8 | 24.2 | 23.8 | |
| 2-stage TranscribeTrain set=L, Prompting method=2-stage Transcribe2026.06 | 9.9 | 25 | 24.5 | |
| TranscribeTrain set=M, Prompting method=Transcribe2026.06 | 10.2 | 27.2 | 26.4 | |
| Transcribe w/ ContextTrain set=L, Prompting method=Transcribe w/ Context2026.06 | 10.4 | 24.8 | 24.7 | |
| Transcribe w/ ContextTrain set=S, Prompting method=Transcribe w/ Context2026.06 | 11 | 27.9 | 26.9 | |
| 2-stage Reason (ours)Train set=S, Prompting method=2-stage Reason (ours)2026.06 | 11 | 26.3 | 26.1 | |
| TranscribeTrain set=L, Prompting method=Transcribe2026.06 | 11.1 | 28.3 | 27.2 | |
| 2-stage TranscribeTrain set=S, Prompting method=2-stage Transcribe2026.06 | 11.5 | 28.2 | 27.4 | |
| TranscribeTrain set=S, Prompting method=Transcribe2026.06 | 11.9 | 31.2 | 29.6 | |
| TranscribeTrain set=-, Prompting method=Transcribe2026.06 | 13.1 | 30 | 28.9 | |
| Transcribe w/ ContextTrain set=-, Prompting method=Transcribe w/ Context2026.06 | 257.6 | 69.2 | 68.3 |