Audio Deepfake Detection on ASVspoof LA and DF 2021
1.49EER (DF)Nes2Net-X
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Nes2Net-XPublication=TIFS 20252026.02 | 1.49 | 1.88 | 1.69 | — | — | |
| OursPublication=Ours2026.02 | 1.87 | 1.15 | 1.51 | — | — | |
| XLSR-MambaPublication=SPL 2025, Augmentation=RawBoost2026.02 | 1.88 | 1.68 | 1.78 | — | — | |
| WaveSpecPublication=ICASSP 20252026.02 | 1.9 | — | — | — | — | |
| SLSPublication=ACMMM 20242026.02 | 1.92 | 2.87 | 2.4 | — | — | |
| TCMPublication=Interspeech 2024, Augmentation=RawBoost2026.02 | 2.06 | 1.03 | 1.55 | — | — | |
| OC-ACSPublication=Interspeech 20242026.02 | 2.19 | 1.3 | 1.75 | — | — | |
| OCKDPublication=ICASSP 20242026.02 | 2.27 | 0.9 | 1.59 | — | — | |
| LSR+LSAPublication=ICASSP 20252026.02 | 2.43 | 1.19 | 1.81 | — | — | |
| MoEPublication=ICASSP 20252026.02 | 2.54 | 2.96 | 2.75 | — | — | |
| MFAPublication=ICASSP 20242026.02 | 2.56 | 5.08 | 3.82 | — | — | |
| WavLM+MFATrain=19LA, Aug=Yes, Fusion=Single2026.06 | 2.56 | 5.08 | 3.82 | 5.08 | 2.52 | |
| AASIST2Publication=ICASSP 20242026.02 | 2.77 | 1.61 | 2.19 | — | — | |
| FTDKDPublication=TASLP 20242026.02 | 2.82 | 2.96 | 2.89 | — | — | |
| Wav2vec+AASISTPublication=Odyssey 2022, Augmentation=RawBoost2026.02 | 2.85 | 0.82 | 1.84 | — | — | |
| AttM-LSTMPublication=Interspeech 20242026.02 | 3.19 | 3.5 | 3.35 | — | — | |
| MFPT (Ours)Train=19LA, Aug=No, Fusion=Single2026.06 | 4.04 | 3.88 | 3.96 | 4.04 | 0.16 | |
| SLIMPublication=NeurIPS 20242026.02 | 4.4 | — | — | — | — | |
| ASP+MLPPublication=Interspeech 20242026.02 | 4.47 | 3.31 | 3.89 | — | — | |
| WavLM+ASP (SCL+CE)Train=19LA, Aug=No, Fusion=Single2026.06 | 4.47 | 3.31 | 3.89 | 4.47 | 1.16 | |
| Donas et al.Train=19LA, Aug=Yes, Fusion=Single2026.06 | 4.98 | 3.54 | 4.26 | 4.98 | 1.44 | |
| Wav2vec2-XLSRTrain=19LA, Aug=Yes, Fusion=Single2026.06 | 5.44 | 7.18 | 6.31 | 7.18 | 1.74 |