SAFE Audio Challenge Task 3: leaderboard
SAFE Audio Challenge Task 3 evaluates generated audio detection systems on the third public SAFE challenge task split.
Metric: Balanced Accuracy (%). Source: huggingface.co. 11 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | baseline-1 | 68.92 | |
| 2 | JAIST-HIS | 65.85 | |
| 3 | ISSF | 65.82 | |
| 4 | ISPL | 64.47 | |
| 5 | baseline-2 | 62.2 | |
| 6 | Anon_Peking | 60.9 | |
| 7 | viper-purdue | 58.11 | |
| 8 | gylin | 50.92 | |
| 9 | safe-test | 50.5 | |
| 10 | DMF | 48.07 |
Interactive version: theaggregate.ai/benchmark?slug=safe-audio-challenge-task-3 · How It Works · Data refreshed daily, snapshot 2026-10-20.