# Dual-Branch Gated Fusion for Open-Set Audio Deepfake Source Tracing

arXiv: https://arxiv.org/abs/2606.10223 · HTML: https://mlsanity.com/paper/2606.10223/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Builds on 22 of its 37 references that are in the map.
- First posted on arXiv 8 Jun 2026.

## Builds on (most cited first, 22 of 22)

- [Energy-based Out-of-distribution Detection](https://mlsanity.com/paper/2010.03759/) (2020)
- [XLS-R: Self-supervised Cross-lingual Speech Representation Learning at Scale](https://mlsanity.com/paper/2111.09296/) (2021)
- [MUSAN: A Music, Speech, and Noise Corpus](https://mlsanity.com/paper/1510.08484/) (2015)
- [Automatic speaker verification spoofing and deepfake detection using wav2vec 2.0 and data augmentation](https://mlsanity.com/paper/2202.12233/) (2022)
- [ADD 2023: the Second Audio Deepfake Detection Challenge](https://mlsanity.com/paper/2305.13774/) (2023)
- [Source Tracing of Audio Deepfake Systems](https://mlsanity.com/paper/2407.08016/) (2024)
- [Where are we in audio deepfake detection? A systematic analysis over generative and detection models](https://mlsanity.com/paper/2410.04324/) (2024)
- [Frame-to-Utterance Convergence: A Spectra-Temporal Approach for Unified Spoofing Detection](https://mlsanity.com/paper/2309.09837/) (2023)
- [Transferable Adversarial Attacks on Audio Deepfake Detection](https://mlsanity.com/paper/2501.11902/) (2025)
- [Multilingual Source Tracing of Speech Deepfakes: A First Benchmark](https://mlsanity.com/paper/2508.04143/) (2025)
- [Adversarial Attacks on Audio Deepfake Detection: A Benchmark and Comparative Study](https://mlsanity.com/paper/2509.07132/) (2025)
- [Attacker Attribution of Audio Deepfakes](https://mlsanity.com/paper/2203.15563/) (2022)
- [Distinguishing Neural Speech Synthesis Models Through Fingerprints in Speech Waveforms](https://mlsanity.com/paper/2309.06780/) (2023)
- [Interpretable All-Type Audio Deepfake Detection with Audio LLMs via Frequency-Time Reinforcement Learning](https://mlsanity.com/paper/2601.02983/) (2026)
- [Securing Voice Biometrics: One-Shot Learning Approach for Audio Deepfake Detection](https://mlsanity.com/paper/2310.03856/) (2023)
- [Generalized Source Tracing: Detecting Novel Audio Deepfake Algorithm with Real Emphasis and Fake Dispersion Strategy](https://mlsanity.com/paper/2406.03240/) (2024)
- [Neural Codec Source Tracing: Toward Comprehensive Attribution in Open-Set Condition](https://mlsanity.com/paper/2501.06514/) (2025)
- [Unveiling Audio Deepfake Origins: A Deep Metric learning And Conformer Network Approach With Ensemble Fusion](https://mlsanity.com/paper/2506.02085/) (2025)
- [TRACE: Training-Free Partial Audio Deepfake Detection via Embedding Trajectory Analysis of Speech Foundation Models](https://mlsanity.com/paper/2604.01083/) (2026)
- [Investigating Prosodic Signatures via Speech Pre-Trained Models for Audio Deepfake Source Attribution](https://mlsanity.com/paper/2412.17796/) (2024)
- [Synthetic Speech Source Tracing using Metric Learning](https://mlsanity.com/paper/2506.02590/) (2025)
- [Open-Set Source Tracing of Audio Deepfake Systems](https://mlsanity.com/paper/2507.06470/) (2025)

## Built on by (most cited first, 2 of 2)

- [Do Transformations Reveal the Truth? Generative Residual Learning for Generalized AI-Generated Image Detection](https://mlsanity.com/paper/2607.08674/) (2026)
- [Uncertainty-Aware Deepfake Detection via Multi-View Structural Learning](https://mlsanity.com/paper/2607.28769/) (2026)

## Abstract (arXiv)

Attributing a synthetic utterance to its originating system remains an open challenge: closed-set models fail to reject unseen synthesizers and produce overconfident predictions. To address this, we propose a dual-branch gated fusion framework that pairs XLSR-53 with CORES, a 66-dimensional descriptor that, unlike prior Linear Filter Bank (LFB)-only work, spans cepstral, oscillatory, rhythmic, energy, and spectral dimensions to capture complementary synthesis artifacts. Our analysis shows XLSR-53 remains discriminative in-domain (ID) while CORES generalizes stably under distribution shift (OOD), yet their naive concatenation fails due to SSL representational imbalance. To resolve this, an input-conditioned gate adaptively weights each branch under joint training with cross-entropy, an energy margin loss for ID/OOD separation, and a gate diversity term. On the MLAAD benchmark, our system achieves 97.6% ID accuracy, 4.9% EERc, and an 83.5% relative FPR95 reduction over the Interspeech 2025 baseline.