Emphasis Detection on EmphAssess (test)
55.11PrecisionSPEECHCOMBINE
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| SPEECHCOMBINEGroup=Group B, Reasoning prompt=Based on the prosody2026.07 | 55.11 | 67.89 | 60.84 | |
| GPT-4O-AUDIOGroup=Topline2026.07 | 33 | 61.68 | 42.99 | |
| FUN-AUDIO-CHATGroup=Group B2026.07 | 23.46 | 37.15 | 28.76 | |
| OSUM-ECHATGroup=Group B2026.07 | 22.51 | 9.02 | 12.88 | |
| STEP-AUDIO2-MINI-THINKGroup=Group B2026.07 | 20.37 | 39.45 | 26.87 | |
| KIMI-AUDIOGroup=Group B2026.07 | 19.71 | 20.04 | 19.88 | |
| STEP-AUDIO2-MINIGroup=Group B2026.07 | 19.59 | 25.76 | 22.25 | |
| AUDIO-FLAMINGOGroup=Group B2026.07 | 17.09 | 33.32 | 22.59 | |
| ASR + TEXT LLMGroup=Group A2026.07 | 15.79 | 26.94 | 19.91 | |
| GLM-4-VOICEGroup=Group B2026.07 | 11.92 | 29.52 | 16.98 | |
| QWEN-2.5-OMNIGroup=Group B2026.07 | 3.75 | 26.26 | 6.56 | |
| CONT. PRE-TRAIN + SFTGroup=Group A2026.07 | 2.12 | 3.42 | 2.61 | |
| CONT. PRE-TRAINGroup=Group A2026.07 | 1.53 | 6.51 | 2.48 |