mlsanityData to 7 Oct 2026

Do Compact SSL Backbones Matter for Audio Deepfake Detection? A Controlled Study with RAPTOR

arXiv 2603.06164 ↗First posted 6 Mar 2026cs.SD cs.AI cs.CLcited by 4

Not on the map. Why?

Abstract (arXiv)

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.

Key facts. Cited by 4 papers in the map (data through 7 Oct 2026). Builds on 42 of its 54 references that are in the map. First posted on arXiv 6 Mar 2026.

Builds on

42 of 54 references.

References · 42

  1. 2017Attention Is All You Needcited by 59,932
  2. 2015Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learningcited by 4,416
  3. 2016Simple and Scalable Predictive Uncertainty Estimation using Deep Ensemblescited by 3,888
  4. 2017What Uncertainties Do We Need in Bayesian Deep Learning for Computer Vision?cited by 2,335
  5. 2021HuBERT: Self-Supervised Speech Representation Learning by Masked Prediction of Hidden Unitscited by 2,301
  6. 2021WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech Processingcited by 1,609
  7. 2019Unsupervised Data Augmentation for Consistency Trainingcited by 1,057
  8. 2017Virtual Adversarial Training: A Regularization Method for Supervised and Semi-Supervised Learningcited by 918
  9. 2021XLS-R: Self-supervised Cross-lingual Speech Representation Learning at Scalecited by 589
  10. 2015MUSAN: A Music, Speech, and Noise Corpuscited by 400
  11. 2021MEMO: Test Time Robustness via Adaptation and Augmentationcited by 376
  12. 2018Aleatoric uncertainty estimation with test-time augmentation for medical image segmentation with convolutional neural networkscited by 278
30 more
  1. 2019ASVspoof 2019: Future Horizons in Spoofed and Fake Audio Detectioncited by 173
  2. 2021AASIST: Audio Anti-Spoofing using Integrated Spectro-Temporal Graph Attention Networkscited by 173
  3. 2022Does Audio Deepfake Detection Generalize?cited by 129
  4. 2022ASVspoof 2021: Towards Spoofed and Deepfake Speech Detection in the Wildcited by 118
  5. 2020An Overview of Voice Conversion and its Challenges: From Statistical Modeling to Deep Learningcited by 111
  6. 2022Automatic speaker verification spoofing and deepfake detection using wav2vec 2.0 and data augmentationcited by 110
  7. 2022ADD 2022: the First Audio Deep Synthesis Detection Challengecited by 94
  8. 2024Recent Advances in Speech Language Models: A Surveycited by 85
  9. 2021RawBoost: A Raw Data Boosting and Augmentation Method applied to Automatic Speaker Verification Anti-Spoofingcited by 64
  10. 2023ADD 2023: the Second Audio Deepfake Detection Challengecited by 64
  11. 2024mHuBERT-147: A Compact Multilingual HuBERT Modelcited by 64
  12. 2024MLAAD: The Multi-Language Audio Anti-Spoofing Datasetcited by 55
  13. 2023AI-Synthesized Voice Detection Using Neural Vocoder Artifactscited by 41
  14. 2024The Codecfake Dataset and Countermeasures for the Universally Detection of Deepfake Audiocited by 31
  15. 2024XLSR-Mamba: A Dual-Column Bidirectional State Space Model for Spoofing Attack Detectioncited by 27
  16. 2021A Survey on Epistemic (Model) Uncertainty in Supervised Learning: Recent Advances and Applicationscited by 24
  17. 2024Temporal-Channel Modeling in Multi-head Self-Attention for Synthetic Speech Detectioncited by 24
  18. 2023Towards generalisable and calibrated synthetic speech detection with self-supervised representationscited by 23
  19. 2023Bayesian Federated Learning: A Surveycited by 22
  20. 2024SpoofCeleb: Speech Deepfake Detection and SASV In The Wildcited by 22
  21. 2024DFADD: The Diffusion and Flow-Matching Based Audio Deepfake Datasetcited by 20
  22. 2025Speech DF Arena: A Leaderboard for Speech DeepFake Detection Modelscited by 20
  23. 2024CtrSVDD: A Benchmark Dataset and Baseline Analysis for Controlled Singing Voice Deepfake Detectioncited by 16
  24. 2020How Far Are We from Robust Voice Conversion: A Surveycited by 14
  25. 2025Comprehensive Layer-wise Analysis of SSL Models for Audio Deepfake Detectioncited by 12
  26. 2025EnvSDD: Benchmarking Environmental Sound Deepfake Detectioncited by 11
  27. 2024Harder or Different? Understanding Generalization of Audio Deepfake Detectioncited by 9
  28. 2025BiCrossMamba-ST: Speech Deepfake Detection with Bidirectional Mamba Spectro-Temporal Cross-Attentioncited by 9
  29. 2025Children's Voice Privacy: First Steps And Emerging Challengescited by 5
  30. 2025Unveiling Audio Deepfake Origins: A Deep Metric learning And Conformer Network Approach With Ensemble Fusioncited by 5

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 4.

4 papers.

  1. 2026Teffic-Audio: Tell Fact from Fictioncited by 1
  2. 2026MADBench: A Benchmark for Modality-Aware Audio Deepfake Detectioncited by 0
  3. 2026Robust Workflow Generation via Adversarial Learning for Audio Deepfake Detectioncited by 0
  4. 2026VietPrism: A large-scale Vietnamese speech and deepfake corpus with diverse dialects and code-switchingcited by 0