Speech Deepfake Detection on CodecFake (CF) unseen (test)
0.9806AccuracyGARUDA-FT
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| GARUDA-FTtraining_mode=fine-tuning2026.06 | 0.9806 | 0.0223 | |
| GARUDAtraining_mode=projection-only2026.06 | 0.946 | 0.0571 | |
| MiO2026.06 | 0.9344 | 0.0777 | |
| Qwen2-Audio-Base-FTtraining_mode=fine-tuning2026.06 | 0.9326 | 0.0542 |