Audio Deepfake Detection on ASVspoof LA 2019
0.35EERQAF-Static
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| QAF-Staticcodec=fine-tuned, SSL backbone=frozen, variant=Method 22026.03 | 0.35 | — | — | — | 46.2 | |
| MoLEx2026.03 | 0.44 | — | — | — | 32.3 | |
| QAF-Staticcodec=frozen, SSL backbone=frozen, variant=Method 22026.03 | 0.44 | — | — | — | 32.3 | |
| QAF-Staticcodec=frozen, SSL backbone=frozen, variant=Method 12026.03 | 0.53 | — | — | — | 18.4 | |
| AttM2026.03 | 0.65 | — | — | — | — | |
| W2V-Large1-LLGF2026.03 | 0.86 | — | — | — | — | |
| W2V2-MusicDETFront-end=W2V2 (Frozen), Back-end=MusicDET2026.05 | 0.86 | — | — | — | — | |
| W2V2-large-DARTS2026.03 | 1.08 | — | — | — | — | |
| W2V2-base-DARTS2026.03 | 1.19 | — | — | — | — | |
| W2V2-AASISTFront-end=W2V2 (Frozen), Back-end=AASIST2026.05 | 1.28 | — | — | — | — | |
| AASISTBack-end=AASIST2026.05 | 1.48 | — | — | — | — | |
| W2V2-ResNetFront-end=W2V2 (Frozen), Back-end=ResNet2026.05 | 2.07 | — | — | — | — | |
| WavLM-AASISTFront-end=WavLM (Frozen), Back-end=AASIST2026.05 | 2.49 | — | — | — | — | |
| W2V-XLSR-LLGF2026.03 | 2.8 | — | — | — | — | |
| H-LoRASamples=2002026.03 | 3.33 | 96.73 | 96.77 | 99.43 | — | |
| HuBERT-XL2026.03 | 3.55 | — | — | — | — | |
| W2V2-LCNNFront-end=W2V2 (Frozen), Back-end=LCNN2026.05 | 3.63 | — | — | — | — | |
| LORASamples=2002026.03 | 4.31 | 95.54 | 95.67 | 99.04 | — | |
| MERT-AASISTFront-end=MERT (Frozen), Back-end=AASIST2026.05 | 4.8 | — | — | — | — | |
| H-LoRASamples=1002026.03 | 5.48 | 94.37 | 94.58 | 98.57 | — | |
| Spec-ResNetFront-end=Spectrogram, Back-end=ResNet2026.05 | 5.58 | — | — | — | — | |
| LORASamples=1002026.03 | 7.11 | 89.39 | 90.4 | 97.45 | — | |
| H-LoRASamples=502026.03 | 8.71 | 88.32 | 89.33 | 96.52 | — | |
| XLS-R+STCA+LMDC2026.02 | 9 | — | — | — | — | |
| WHISPERSamples=2002026.03 | 9.57 | 89.93 | 90.69 | 95.82 | — | |
| WHISPERSamples=1002026.03 | 12.6 | 87.42 | 88.19 | 93.71 | — | |
| LORASamples=502026.03 | 13.58 | 78.89 | 80.34 | 92.86 | — | |
| XLSR+AASIST22026.02 | 15 | — | — | — | — | |
| WHISPERSamples=502026.03 | 16.16 | 72.54 | 72.99 | 90.85 | — | |
| XLSR+Conformer+TCMParams (M)=3192026.02 | 19 | — | — | — | — | |
| Wav2Vec2+AASISTParams (M)=317.842026.02 | 22.1 | — | — | — | — | |
| HyperPotterParams (M)=317.872026.02 | 23 | — | — | — | — | |
| XLSR+SLSParams (M)=3402026.02 | 23.1 | — | — | — | — | |
| Wav2Vec2+AASIST(Baseline)Params (M)=317.842026.02 | 26 | — | — | — | — | |
| WavLM+MFA2026.02 | 42 | — | — | — | — | |
| XLSR+MambaParams (M)=3192026.02 | 42.1 | — | — | — | — | |
| XLSR+BiCrossMambaParams (M)=318.212026.02 | 71 | — | — | — | — | |
| XLSR+MoE2026.02 | 74 | — | — | — | — | |
| AASISTModel Input=Audio2026.01 | — | — | 98.03 | — | — | |
| AASIST-LModel Input=Audio2026.01 | — | — | 97.06 | — | — | |
| ALLM4ADDModel Input=Audio2026.01 | — | — | 99.39 | — | — | |
| AMSDFModel Input=wav2vec2.02026.01 | — | — | 96.88 | — | — | |
| BaselineModel Input=Audio, Backbone=Qwen2.5-Omni-3B2026.01 | — | — | 98.76 | — | — | |
| DFALLMModel Input=Audio2026.01 | — | — | 99.15 | — | — | |
| FGModel Input=Hubert2026.01 | — | — | 95.22 | — | — | |
| HyperPotterParams (M)=317.872026.02 | — | 98.9 | — | — | — | |
| RawGAT-STModel Input=Audio2026.01 | — | — | 98.14 | — | — | |
| RawNet2Model Input=Audio2026.01 | — | — | 93.12 | — | — | |
| SDD-APALLMModel Input=Audio + CQT, Backbone=Qwen2.5-Omni-3B2026.01 | — | — | 99.46 | — | — | |
| TSSDNetModel Input=Audio2026.01 | — | — | 89.93 | — | — | |
| Wav2Vec2+AASISTParams (M)=317.842026.02 | — | 98.9 | — | — | — | |
| Wav2Vec2+AASIST (Baseline)Params (M)=317.842026.02 | — | 98.8 | — | — | — | |
| XLSR+BiCrossMambaParams (M)=318.212026.02 | — | 96.7 | — | — | — | |
| XLSR+Conformer+TCMParams (M)=3192026.02 | — | 99.1 | — | — | — | |
| XLSR+MambaParams (M)=3192026.02 | — | 98 | — | — | — | |
| XLSR+SLSParams (M)=3402026.02 | — | 98.9 | — | — | — |