Do Compact SSL Backbones Matter for Audio Deepfake Detection? A Controlled Study with RAPTOR
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2026
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866917318261473280 |
|---|---|
| author | Kulkarni, Ajinkya Dowerah, Sandipana Kulkarni, Atharva Alumäe, Tanel Doss, Mathew Magimai |
| author_facet | Kulkarni, Ajinkya Dowerah, Sandipana Kulkarni, Atharva Alumäe, Tanel Doss, Mathew Magimai |
| contents | 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. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_06164 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Do Compact SSL Backbones Matter for Audio Deepfake Detection? A Controlled Study with RAPTOR Kulkarni, Ajinkya Dowerah, Sandipana Kulkarni, Atharva Alumäe, Tanel Doss, Mathew Magimai Sound Artificial Intelligence Computation and Language 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. |
| title | Do Compact SSL Backbones Matter for Audio Deepfake Detection? A Controlled Study with RAPTOR |
| topic | Sound Artificial Intelligence Computation and Language |
| url | https://arxiv.org/abs/2603.06164 |