Audio Tagging on AudioSet (test)
50mAPSOTA
Evaluation Results
| Method | Links | |
|---|---|---|
| SOTAtype=non-CLAP-based2026.01 | 50 | |
| EAT-CSUCategory=Proposed, # Parameters=111 M2026.05 | 49.61 | |
| EAT-CHCategory=Proposed, # Parameters=111 M2026.05 | 49.02 | |
| M2D2type=CLAP-based, protocol=linear classification layer fine-tuning2026.01 | 49 | |
| Efficient Audio Transformer (EAT)Category=Self-Supervised Learning, # Parameters=88 M2026.05 | 48.6 | |
| AST (Ensemble-M)e2e=false, Params=526.6 * 10^62021.06 | 48.5 | |
| BEATsCategory=Self-Supervised Learning, # Parameters=90 M2026.05 | 48 | |
| SLAPtype=CLAP-based, protocol=linear classification layer fine-tuning2026.01 | 47.8 | |
| Audio Masked Autoencoder (AudioMAE)Category=Self-Supervised Learning, # Parameters=86 M2026.05 | 47.3 | |
| Masked Spectrogram Prediction (MaskSpec)Category=Self-Supervised Learning, # Parameters=86 M2026.05 | 47.1 | |
| ASTe2e=false, Params=88.1 * 10^62021.06 | 45.9 | |
| Audio Spectrogram Transformer (AST)Category=Supervised Learning, # Parameters=86 M2026.05 | 45.9 | |
| ERANN-1-6e2e=false, Params=54.5 * 10^62021.06 | 45 | |
| ERANN-2-5e2e=false, Params=38.2 * 10^62021.06 | 44.6 | |
| Multimodal Bottleneck Transformer (MBT)Category=Supervised Learning, # Parameters=86 M2026.05 | 44.3 | |
| BLATtype=CLAP-based, protocol=linear classification layer fine-tuning2026.01 | 44 | |
| Wavegram-Logmel-CNNe2e=false, Params=81.1 * 10^62021.06 | 43.9 | |
| CNN14e2e=false, Params=80.7 * 10^62021.06 | 43.1 | |
| Pretrained Audio Neural Networks (PANNs)Category=Supervised Learning, # Parameters=81 M2026.05 | 43.1 | |
| Conformer-based Self-Supervised LearningCategory=Self-Supervised Learning, # Parameters=88 M2026.05 | 41.1 | |
| Wavegram-CNNe2e=true, Params=81.0 * 10^62021.06 | 38.9 | |
| AemNet-DW (WM=2)e2e=true, Params=3.0 * 10^62021.06 | 34 | |
| AemNet-DW (WM=1)e2e=true, Params=1.2 * 10^62021.06 | 32.9 | |
| ESResNeXte2e=false, Params=31.1 * 10^62021.06 | 28.2 | |
| AudioCLIPCategory=Supervised Learning, # Parameters=93 M2026.05 | 25.9 |