Do Compact SSL Backbones Matter for Audio Deepfake Detection? A Controlled Study with RAPTOR
Abstract
Self-supervised learning (SSL) underpins modern audio deepfake detection, yet most prior work centers on a single large wav2vec2-XLSR backbone, leaving compact under studied. We present RAPTOR, Representation Aware Pairwise-gated Transformer for Out-of-domain Recognition a controlled study of compact SSL backbones from the HuBERT and WavLM within a unified pairwise-gated fusion detector, evaluated across 14 cross-domain benchmarks. We show that multilingual HuBERT pre-training is the primary driver of cross-domain robustness, enabling 100M models to match larger and commercial systems. Beyond EER, we introduce a test-time augmentation protocol with perturbation-based aleatoric uncertainty to expose calibration differences invisible to standard metrics: WavLM variants exhibit overconfident miscalibration under perturbation, whereas iterative mHuBERT remains stable. These findings indicate that SSL pre-training trajectory, not model scale, drives reliable audio deepfake detection.
Cite
@article{arxiv.2603.06164,
title = {Do Compact SSL Backbones Matter for Audio Deepfake Detection? A Controlled Study with RAPTOR},
author = {Ajinkya Kulkarni and Sandipana Dowerah and Atharva Kulkarni and Tanel Alumäe and Mathew Magimai Doss},
journal= {arXiv preprint arXiv:2603.06164},
year = {2026}
}
Comments
Submitted to Interspeech 2026, 4 pages, 2 figures