mlsanityData to 14 Oct 2026

Domain-Adaptive Dual-Gating Mixture of Experts for Generalizable Speech Deepfake Detection

arXiv 2609.33709 ↗First posted 27 Sep 2026eess.AS cs.SDcited by 0

On the map: Mixture-of-experts → Multimodal MoE models

Abstract (arXiv)

Recent advances in speech deepfake detection (SDD) have leveraged the Mixture of Experts (MoE) to enhance generalization capacity. However, existing gating networks often overlook the acoustic and temporal cues of deepfakes. In this work, we propose a novel domain-adaptive dual-gating MoE (DADGMoE) framework for SDD under unseen attack types and acoustic conditions. Our innovative dual-gating mechanism leverages Sinc-layer-based filters to process both low-level acoustic signals (raw waveforms) and high-level speech representations from a large self-supervised learning (SSL) model. It further incorporates domain prototypes to guide expert routing based on implicit deepfake patterns. The lightweight affine experts process the routed inputs. Experiments show that our DADGMoE significantly outperforms the baseline, achieving up to a 40.8% relative EER reduction on challenging out-of-dataset benchmarks. This framework demonstrates superior generalization capabilities and efficient design.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Multimodal MoE models, one of 12 directions in Mixture-of-experts. Builds on 28 of its 32 references that are in the map. First posted on arXiv 27 Sep 2026.

Builds on

28 of 32 references.

Closest ancestors · 3

Same direction, newest first

  1. 2025Wav2DF-TSL: Two-stage Learning with Efficient Pre-training and Hierarchical Experts Fusion for Robust Audio Deepfake Detection4 Sep
  2. 2024Leveraging Mixture of Experts for Improved Speech Deepfake Detection24 Sep
  3. 2024Mixture of Experts Fusion for Fake Audio Detection Using Frozen wav2vec 2.018 Sep

Other references · 25

  1. 2014Adam: A Method for Stochastic Optimizationcited by 37,240
  2. 2021XLS-R: Self-supervised Cross-lingual Speech Representation Learning at Scalecited by 601
  3. 2024A Survey on Mixture of Experts in Large Language Modelscited by 237
  4. 2019Domain-Specific Batch Normalization for Unsupervised Domain Adaptationcited by 197
  5. 2018Speaker Recognition from Raw Waveform with SincNetcited by 166
  6. 2020End-to-end anti-spoofing with RawNet2cited by 164
  7. 2021ASVspoof 2021: accelerating progress in spoofed and deepfake speech detectioncited by 161
  8. 2022Does Audio Deepfake Detection Generalize?cited by 135
  9. 2022Automatic speaker verification spoofing and deepfake detection using wav2vec 2.0 and data augmentationcited by 117
  10. 2022ADD 2022: the First Audio Deep Synthesis Detection Challengecited by 96
  11. 2024ASVspoof 5: Crowdsourced Speech Data, Deepfakes, and Adversarial Attacks at Scalecited by 91
  12. 2021RawBoost: A Raw Data Boosting and Augmentation Method applied to Automatic Speaker Verification Anti-Spoofingcited by 67
13 more
  1. 2023ADD 2023: the Second Audio Deepfake Detection Challengecited by 64
  2. 2021ASVspoof 2019: spoofing countermeasures for the detection of synthesized, converted and replayed speechcited by 58
  3. 2023Test-Time Domain Adaptation by Learning Domain-Aware Batch Normalizationcited by 31
  4. 2016Attentive Pooling Networkscited by 28
  5. 2024Temporal-Channel Modeling in Multi-head Self-Attention for Synthetic Speech Detectioncited by 25
  6. 2025Speech DF Arena: A Leaderboard for Speech DeepFake Detection Modelscited by 21
  7. 2025Nes2Net: A Lightweight Nested Architecture for Foundation Model Driven Speech Anti-spoofingcited by 19
  8. 2024AASIST3: KAN-Enhanced AASIST Speech Deepfake Detection using SSL Features and Additional Regularization for the ASVspoof 2024 Challengecited by 9
  9. 2025From Sharpness to Better Generalization for Speech Deepfake Detectioncited by 5
  10. 2025MoLEx: Mixture of LoRA Experts in Speech Self-Supervised Models for Audio Deepfake Detectioncited by 5
  11. 2025Mixture of Low-Rank Adapter Experts in Generalizable Audio Deepfake Detectioncited by 2
  12. 2025Addressing Gradient Misalignment in Data-Augmented Training for Robust Speech Deepfake Detectioncited by 2
  13. 2024What do neural networks listen to? Exploring the crucial bands in Speech Enhancement using Sinc-convolutioncited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2025Attention-based Mixture of Experts for Robust Speech Deepfake Detection
  2. 2026Leveraging Gradient Reversal Loss and Multitask Learning for Datasets-Aware Audio Deepfake Detection
  3. 2026DF-MoE: Generalizable Deepfake Detection via Multimodal Sparse Mixture-of-Experts
  4. 2026Task-Aware Joint Pruning and Distillation for Efficient Audio Deepfake Detection
  5. 2026Teacher-Student Structure for Domain Adaptation in Ensemble Audio-Visual Video Deepfake Detection
  6. 2025WaveSP-Net: Learnable Wavelet-Domain Sparse Prompt Tuning for Speech Deepfake Detection
  7. 2026Interpretable All-Type Audio Deepfake Detection with Audio LLMs via Frequency-Time Reinforcement Learning
  8. 2026A Multi-Domain Feature Fusion Framework for Generalizable Deepfake Detection Across Different Generators