Musicality Evaluation on Music Arena
0.7388AccuracySongEval-RM
Evaluation Results
| Method | Links | |
|---|---|---|
| SongEval-RM2026.02 | 0.7388 | |
| CMI-RMfine-tuned=true, training_variant=CMI + MusicEval2026.02 | 0.7343 | |
| CMI-RMfine-tuned=true, training_variant=CMI-Pref2026.02 | 0.7141 | |
| Gemini2.5-pro2026.02 | 0.6975 | |
| Gemini3-pro2026.02 | 0.6885 | |
| audiobox-CU2026.02 | 0.6776 | |
| audiobox-PQ2026.02 | 0.6754 | |
| audiobox-CE2026.02 | 0.6425 | |
| CMI-RMfine-tuned=false, training_variant=Distill only2026.02 | 0.6425 | |
| Gemini2.5-flash2026.02 | 0.6412 | |
| PAM score2026.02 | 0.6313 | |
| Qwen3-omni2026.02 | 0.5963 | |
| audiobox-PC2026.02 | 0.5873 | |
| Qwen2.5-omni2026.02 | 0.3605 | |
| Qwen2-audio2026.02 | 0.0599 |