Deepfake audio detection with spectral features and ResNeXt-based architecture

Gul Tahaoglu, Daniele Baracchi, Dasara Shullani, Massimo Iuliani, Alessandro Piva

Abstract: Highlights•A novel ResNeXt-based architecture for synthetic audio detection.•Audio embeddings leverage LFCC, MFCC, and CQCC spectral features.•Demonstrates robustness across ASVspoof 2019 PA, LA, and 2021 LA datasets.

External IDs:dblp:journals/kbs/TahaogluBSIP25