Audio Watermarking on VCTK cross-dataset transfer
100Detection Accuracy (Accdet)LambdaMark
Evaluation results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| LambdaMarkDistortions=Without2026.06 | 100 | 98.13 | 100 | 100 | 100 | 1 | 4.26 | 4.81 | |
| LambdaMarkDistortions=With2026.06 | 99.88 | 95.08 | 99.75 | 100 | 99.88 | 0.999 | 3.74 | 3.65 | |
| AudioMarkNetDistortions=Without2026.06 | 93.5 | 83.54 | 100 | 87 | 93.05 | 0.935 | 4.6 | 4.54 | |
| AudioMarkNetDistortions=With2026.06 | 85.54 | 78.92 | 97.62 | 71.25 | 79.91 | 0.855 | 3.92 | 3.33 |