Utterance-level Audio Spoofing Detection on LlamaPartialSpoof All spoof combined
0.2208EERTRACE
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| TRACEThreshold calibration strategy=PartialSpoof-dev-calibrated, corpus=Full Corpus2026.04 | 0.2208 | 0.839 | |
| TRACEThreshold calibration strategy=subset-optimal, corpus=Full Corpus2026.04 | 0.2412 | 0.839 | |
| PartialSpoof-trained baselineThreshold calibration strategy=subset-optimal2026.04 | 0.2449 | — | |
| w2v2-XLsr-SALThreshold calibration strategy=subset-optimal2026.04 | 0.3552 | 0.553 | |
| WavLM-SALThreshold calibration strategy=subset-optimal2026.04 | 0.366 | 0.561 | |
| WavLM-BAMThreshold calibration strategy=subset-optimal2026.04 | 0.4258 | 0.534 | |
| w2v2-Large-MultiResThreshold calibration strategy=subset-optimal2026.04 | 0.4749 | — |