英文摘要 |
Reliability of Automatic Speaker Veri cation (ASV) systems has always been a concern in dealing with spoo ng attacks. Among these attacks, replay attack is the simplest and the easiest accessible method. This paper describes a replay spoo ng detection system applied to ASVspoof2017 corpus. To reach this goal, features such as Constant-Q Cepstral Coe - cients (CQCC), Modi ed Group Delay (MGD), Mel Frequency Cepstral Coe cients (MFCC), Relative Spectral Perceptual Linear Predictive (RASTA-PLP) and Linear Prediction Cepstral Coe cients (LPCC), and di erent classi ers including Gaussian Mixture Models (GMM), Multi- Layer Perceptron (MLP), Support Vector Machine (SVM) and Linear Gaussian (LG) classi er have been employed. We also used identity vector (i-vector) based utterance representation. Finally, scores of di erent subsystems have been fused to construct the proposed system. The results show that the best performance is attained using this score level fusion. |