Tagging on MTT Magnatagatune (test)
92MTT AUCState-of-the-art
Evaluation results
| Method | Links | ||
|---|---|---|---|
| State-of-the-artEvaluation protocol=Reported performance2021.07 | 92 | 38.4 | |
| Pre-trainedEvaluation protocol=Reported performance2021.07 | 92 | 35.9 | |
| JUKEBOXPre-training type=CALM, Evaluation protocol=Probing2021.07 | 91.5 | 41.4 | |
| Prior work [11]Pre-training=Supervised (S), Evaluation Protocol=Shallow classifier, Training Samples=15,2472021.10 | 91.5 | — | |
| cf_LArchitecture=SS Conformer, Model Scale=Large, Pre-training=Self-supervised (SS), Evaluation Protocol=Fine-tuning, Training Samples=15,2472021.10 | 91.2 | — | |
| From scratchEvaluation protocol=Reported performance2021.07 | 90.7 | 38.4 | |
| MUSICNNPre-training type=Tagging, Evaluation protocol=Probing2021.07 | 90.6 | 38.3 | |
| cf_SArchitecture=SS Conformer, Model Scale=Small, Pre-training=Self-supervised (SS), Evaluation Protocol=Fine-tuning, Training Samples=15,2472021.10 | 90.2 | — | |
| CHOIPre-training type=Tagging, Evaluation protocol=Probing2021.07 | 89.7 | 36.4 | |
| CLMRPre-training type=Contrastive, Evaluation protocol=Probing2021.07 | 89.4 | 36.1 | |
| Prior work [37]Pre-training=Self-supervised (SS), Evaluation Protocol=Shallow classifier, Training Samples=15,247, Footnote=Logmel only2021.10 | 89.3 | — | |
| MFCCPre-training type=No pre-training, Evaluation protocol=Probing2021.07 | 85.8 | 30.2 | |
| CHROMAPre-training type=No pre-training, Evaluation protocol=Probing2021.07 | 77.6 | 18.5 |