Table 4 Results for synthesized speech and voice conversion
From: DeepDet: YAMNet with BottleNeck Attention Module (BAM) for TTS synthesis detection
Set category | Spoofing system | Accuracy (%) | Precision (%) | Recall (%) | EER (%) | Min-tDCF |
|---|---|---|---|---|---|---|
Eval | VC | 93.3 | 97 | 95 | 0.90 | 0.06 |
TTS | 99.8 | 99.6 | 99.4 | 0.50 | 0.005 | |
LA(overall) | 99.92 | 99.2 | 99.76 | 0.042 | 0.0015 | |
Dev | VC | 94.1 | 95 | 95 | 0.50 | 0.11 |
| Â | TTS | 98.8 | 98.6 | 99.1 | 0.010 | 0.015 |
| Â | LA(overall) | 98.9 | 98.2 | 98.9 | 0.015 | 0.002 |