Security and Communication Networks

Research Article

One-Class Fake Speech Detection Based on Improved Support Vector Data Description

Table 3

The detection performance of different methods on the ASVspoof 2019 dataset.
MethodEER (%)
OC-SVM + CQT23.97
DSVDD + CQT12.52
GMM + CQCC26.37
GMM + LFCC28.12
Res-TSSDNet + PCM40.19
Res-TSSDNet + CQT19.97
ResNet + spectrum22.33
ResNet + MFCC22.59
Proposed (AESVDD-1 + CQT)8.10
Proposed (AESVDD-2 + CQT)9.20
Bold values indicate the optimal results.