AudioSet
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
AudioSet SingleSource (test)
4.26SDR
5
AudioSet
4.35FAD
4
AudioSet
3.26FAD
4
AudioSet dynamic single-source moving clips
13.7DoA MAE
4
Audioset
63.7ALOHa Score
4
AudioSet
3.32Perplexity (PPL0)
4
AudioSet AS-A
49.6mAP
4
AudioSet 3Mix
6.5SNRi
4
AudioSet
0.447LA-CLAP
4
AudioSet (eval)
24.26FD
4
AudioSet Music (eval)
56.3MuLan Similarity (Audio to GT Text)
4
AudioSet Drums 64x64
0.987SSIM (t=16)
4
AudioSet ZSL (test)
72.25mAP (Seen)
4
AudioSet dynamic single-source moving clips
62.3Trajectory Accuracy @ 20°
3
AudioSet dynamic single-source moving clips
62.8mAP
3
AudioSet (val)
68.1Top-10 Accuracy
3
AudioSet (val)
69.8Acc @10
3
AudioSet AS-A Audio-only (test)
0.3mAP
3
AudioSet Strong (test)
81.3AUROC
2
AudioSet
7.6SI-SDR
2
AudioSet Music (eval)
2.24FADVGG
2
AudioSet Stage 2 - Human Generated QA (test)
90Instruction Following Rate
1
AudioSet Stage 1 - GPT-4 Generated QA (test)
87.1Instruction Following Rate
1
AudioSet (test)
—Primary metric
0