Audio Deepfake Detection on xTTS v2
2.24FNRUnconditional Sampling
Evaluation results
| Method | Links | |
|---|---|---|
| Unconditional SamplingDetector Backbone=VIT, Training Dataset=VoxCelebSpoof, Visualization=ConstantQ2026.06 | 2.24 | |
| FoeGlass (Warm Start)Detector Backbone=AST, Training Dataset=ASVspoof52026.06 | 3.97 | |
| Unconditional SamplingDetector Backbone=AST, Training Dataset=ASVspoof52026.06 | 4.24 | |
| FoeGlass (Cold Start)Detector Backbone=AST, Training Dataset=ASVspoof52026.06 | 4.86 | |
| Unconditional SamplingDetector Backbone=VIT, Training Dataset=VoxCelebSpoof, Visualization=MelSpectrogram2026.06 | 8.72 | |
| Unconditional SamplingDetector Backbone=VIT, Training Dataset=VoxCelebSpoof, Visualization=MFCC2026.06 | 9.16 | |
| Unconditional SamplingDetector Backbone=AST, Training Dataset=VoxCelebSpoof2026.06 | 9.68 | |
| FoeGlass (Cold Start)Detector Backbone=VIT, Training Dataset=ASVspoof5, Visualization=MelSpectrogram2026.06 | 12.87 | |
| Unconditional SamplingDetector Backbone=VIT, Training Dataset=ASVspoof5, Visualization=MelSpectrogram2026.06 | 23.08 | |
| FoeGlass (Cold Start)Detector Backbone=AST, Training Dataset=VoxCelebSpoof2026.06 | 48.43 | |
| Unconditional SamplingDetector Backbone=VIT, Training Dataset=ASVspoof5, Visualization=ConstantQ2026.06 | 53.8 | |
| FoeGlass (Warm Start)Detector Backbone=AST, Training Dataset=VoxCelebSpoof2026.06 | 63.3 | |
| FoeGlass (Warm Start)Detector Backbone=VIT, Training Dataset=ASVspoof5, Visualization=MelSpectrogram2026.06 | 68.12 | |
| FoeGlass (Cold Start)Detector Backbone=VIT, Training Dataset=VoxCelebSpoof, Visualization=MFCC2026.06 | 71.6 | |
| FoeGlass (Cold Start)Detector Backbone=VIT, Training Dataset=VoxCelebSpoof, Visualization=ConstantQ2026.06 | 80.72 | |
| FoeGlass (Cold Start)Detector Backbone=VIT, Training Dataset=VoxCelebSpoof, Visualization=MelSpectrogram2026.06 | 87.87 | |
| FoeGlass (Warm Start)Detector Backbone=VIT, Training Dataset=VoxCelebSpoof, Visualization=MelSpectrogram2026.06 | 88.83 | |
| Unconditional SamplingDetector Backbone=VIT, Training Dataset=ASVspoof5, Visualization=MFCC2026.06 | 88.92 | |
| FoeGlass (Cold Start)Detector Backbone=VIT, Training Dataset=ASVspoof5, Visualization=MFCC2026.06 | 91.92 | |
| FoeGlass (Warm Start)Detector Backbone=VIT, Training Dataset=VoxCelebSpoof, Visualization=MFCC2026.06 | 93.13 | |
| FoeGlass (Cold Start)Detector Backbone=VIT, Training Dataset=ASVspoof5, Visualization=ConstantQ2026.06 | 93.63 | |
| FoeGlass (Warm Start)Detector Backbone=VIT, Training Dataset=ASVspoof5, Visualization=ConstantQ2026.06 | 93.76 | |
| FoeGlass (Warm Start)Detector Backbone=VIT, Training Dataset=ASVspoof5, Visualization=MFCC2026.06 | 94 | |
| FoeGlass (Warm Start)Detector Backbone=VIT, Training Dataset=VoxCelebSpoof, Visualization=ConstantQ2026.06 | 96.29 |