Environmental Sound Deepfake Detection on EnvSDD Track 1 Final ranking-phase 1.0
1.6EEREnsemble
Evaluation results
| Method | Links | |
|---|---|---|
| EnsembleEnsemble components=0.4×BEATs-AASIST + 0.6×BEAT2AASIST (Channel, SE-gate, k=10)2025.12 | 1.6 | |
| EnsembleEnsemble components=0.3×BEAT2AASIST (Freq., SE-gate, k=4) + 0.7×BEAT2AASIST (Channel, SE-gate, k=10)2025.12 | 1.62 | |
| BEAT2AASISTFusion strategy=Channel-based, Gate type=SE-gate, k (number of layers)=102025.12 | 1.7 | |
| BEATs-AASIST2025.12 | 1.88 | |
| BEAT2AASISTFusion strategy=Frequency-based, Gate type=SE-gate, k (number of layers)=42025.12 | 1.92 |