Abstract
In this paper, we investigate the properties of the cepstrogram and demonstrate its effectiveness as a powerful feature for countermeasure against replay attacks. Cepstrum analysis of replay attacks suggests that crucial information for anti-spoofing against replay attacks may retain in the cepstrogram. Experimental results on the ASVspoof 2019 physical access (PA) database demonstrate that, compared with other features, the cepstrogram dominates in both single and fusion systems when building countermeasures against replay attacks. Our LCNN-based single and fusion systems with the cepstrogram feature outperform the corresponding LCNN-based systems without using the cepstrogram feature and several state-of-the-art (SOTA) single and fusion systems in the literature.
Abstract (translated)
URL
https://arxiv.org/abs/2204.04333
