Automatic Speech Recognition on MCIF long track
5.88WERCascaded
Evaluation results
| Method | Links | |
|---|---|---|
| CascadedPrompt Evaluation Mode=Fixed2026.06 | 5.88 | |
| Cascaded+FT (Contrastive)Prompt Evaluation Mode=Fixed2026.06 | 5.9 | |
| CascadedPrompt Evaluation Mode=Mixed2026.06 | 6.85 | |
| Cascaded+FT (Contrastive)Prompt Evaluation Mode=Mixed2026.06 | 9.76 | |
| N2+IT+SHASPrompt Evaluation Mode=Mixed2026.06 | 11.25 | |
| N2+IT+SHASPrompt Evaluation Mode=Fixed2026.06 | 11.72 | |
| N2Prompt Evaluation Mode=Fixed2026.06 | 25.98 | |
| Temp. Sampling (T = 2)Prompt Evaluation Mode=Fixed2026.06 | 27.58 | |
| N2+AvgPrompt Evaluation Mode=Fixed2026.06 | 29.36 | |
| N2Prompt Evaluation Mode=Mixed2026.06 | 30.46 | |
| Fixed-Prob. SamplingPrompt Evaluation Mode=Fixed2026.06 | 30.59 | |
| Fixed-Prob. SamplingPrompt Evaluation Mode=Mixed2026.06 | 33.61 | |
| N2+AvgPrompt Evaluation Mode=Mixed2026.06 | 34.48 | |
| Qwen 2.5 OmniPrompt Evaluation Mode=Mixed2026.06 | 35.35 | |
| N2+IT+AvgPrompt Evaluation Mode=Mixed2026.06 | 36.88 | |
| N2+IT+AvgPrompt Evaluation Mode=Fixed2026.06 | 37.65 | |
| Temp. Sampling (T = 2)Prompt Evaluation Mode=Mixed2026.06 | 38.43 | |
| Qwen 2.5 OmniPrompt Evaluation Mode=Fixed2026.06 | 53.4 | |
| Chain-of-Thought (CoT)Prompt Evaluation Mode=Fixed2026.06 | 79.24 | |
| Chain-of-Thought (CoT)Prompt Evaluation Mode=Mixed2026.06 | 80.04 |