Towards Generalisable And Calibrated Synthetic Speech Detection With Self-supervised Representations

Abstract

Generalisation -- the ability of a model to perform well on unseen data -- is crucial for building reliable deepfake detectors. However, recent studies have shown that the current audio deepfake models fall short of this desideratum. In this work we investigate the potential of pretrained self-supervised representations in building general and calibrated audio deepfake detection models. We show that large frozen representations coupled with a simple logistic regression classifier are extremely effective in achieving strong generalisation capabilities: compared to the RawNet2 model, this approach reduces the equal error rate from 30.9% to 8.8% on a benchmark of eight deepfake datasets, while learning less than 2k parameters. Moreover, the proposed method produces considerably more reliable predictions compared to previous approaches making it more suitable for realistic use.

Stats

• citations0
• S2 citations—
• github stars0
• HF likes0
• heat score0.00
• arxiv keypascu2023towards

Related papers

• Investigating Self-supervised Representations For Audio-visual Deepfake Detection (2025)0.00
• Training-free Deepfake Voice Recognition By Leveraging Large-scale Pre-trained Models (2024)9.23
• Characterizing The Temporal Dynamics Of Universal Speech Representations For Generalizable Deepfake Detection (2023)6.77
• Combining Automatic Speaker Verification And Prosody Analysis For Synthetic Speech Detection (2022)10.48
• Adaptive Re-calibration Of Channel-wise Features For Adversarial Audio Classification (2022)0.00
• Anomaly Detection And Localization For Speech Deepfakes Via Feature Pyramid Matching (2025)4.52
• Zero-day Audio Deepfake Detection Via Retrieval Augmentation And Profile Matching (2025)0.00
• Universal Paralinguistic Speech Representations Using Self-supervised Conformers (2021)10.48