Music Captioning on LP-MusicCaps-MC (test)
30.8BLEUMusiLingo
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| MusiLingoTrain Dataset=MSD+MC2026.01 | 30.8 | 21.6 | 21.7 | 86.8 | |
| SLAM-LLMEncoder=MuQ-MuLan, Encoder type=Sequence-wise, Train Dataset=MC2026.01 | 27.9 | 20.5 | 21.1 | 86.9 | |
| SLAM-LLMEncoder=Microsoft-CLAP2023, Encoder type=Sequence-wise, Train Dataset=MC2026.01 | 27.5 | 20.3 | 21.1 | 86.8 | |
| SLAM-LLMEncoder=MuQ, Encoder type=Frame-wise, Train Dataset=MC2026.01 | 27 | 19.2 | 19.1 | 86.8 | |
| SLAM-LLMEncoder=Laion-CLAP, Encoder type=Sequence-wise, Train Dataset=MC2026.01 | 25.9 | 19.6 | 20.9 | 86.6 | |
| SLAM-LLMEncoder=MusicFM, Encoder type=Frame-wise, Train Dataset=MC2026.01 | 25.1 | 18.6 | 19.2 | 86.2 | |
| SLAM-LLMEncoder=MERT, Encoder type=Frame-wise, Train Dataset=MC2026.01 | 22.2 | 11.6 | 16.8 | 86.3 | |
| LP-MusicCapsTrain Dataset=MSD+MTT+MC2026.01 | 14.7 | 22.4 | 21.5 | 87.8 | |
| MusCapsTrain Dataset=Private2026.01 | 10.2 | 17 | 22.2 | 83.5 |