# Towards Trustworthy Audio Deepfake Detection: A Systematic Framework for Diagnosing and Mitigating Gender Bias

arXiv: https://arxiv.org/abs/2605.09087 · HTML: https://mlsanity.com/paper/2605.09087/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 20 of its 34 references that are in the map.
- First posted on arXiv 9 May 2026.

## Builds on (most cited first, 20 of 20)

- [wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations](https://mlsanity.com/paper/2006.11477/) (2020)
- [Equality of Opportunity in Supervised Learning](https://mlsanity.com/paper/1610.02413/) (2016)
- [Unsupervised Domain Adaptation by Backpropagation](https://mlsanity.com/paper/1409.7495/) (2014)
- [A Survey on Bias and Fairness in Machine Learning](https://mlsanity.com/paper/1908.09635/) (2019)
- [WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech Processing](https://mlsanity.com/paper/2110.13900/) (2021)
- [Fairness Without Demographics in Repeated Loss Minimization](https://mlsanity.com/paper/1806.08010/) (2018)
- [Fairness in Criminal Justice Risk Assessments: The State of the Art](https://mlsanity.com/paper/1703.09207/) (2017)
- [AASIST: Audio Anti-Spoofing using Integrated Spectro-Temporal Graph Attention Networks](https://mlsanity.com/paper/2110.01200/) (2021)
- [End-to-end anti-spoofing with RawNet2](https://mlsanity.com/paper/2011.01108/) (2020)
- [ASVspoof 5: Crowdsourced Speech Data, Deepfakes, and Adversarial Attacks at Scale](https://mlsanity.com/paper/2408.08739/) (2024)
- [Preserving Fairness Generalization in Deepfake Detection](https://mlsanity.com/paper/2402.17229/) (2024)
- [Improving Fairness in Deepfake Detection](https://mlsanity.com/paper/2306.16635/) (2023)
- [GBDF: Gender Balanced DeepFake Dataset Towards Fair DeepFake Detection](https://mlsanity.com/paper/2207.10246/) (2022)
- [FairSSD: Understanding Bias in Synthetic Speech Detectors](https://mlsanity.com/paper/2404.10989/) (2024)
- [Real-time Detection of AI-Generated Speech for DeepFake Voice Conversion](https://mlsanity.com/paper/2308.12734/) (2023)
- [PhonemeDF: A Synthetic Speech Dataset for Audio Deepfake Detection and Naturalness Evaluation](https://mlsanity.com/paper/2603.15037/) (2026)
- [Gender Fairness in Audio Deepfake Detection: Performance and Disparity Analysis](https://mlsanity.com/paper/2603.09007/) (2026)
- [Investigating the Impact of Speech Enhancement on Audio Deepfake Detection in Noisy Environments](https://mlsanity.com/paper/2603.14767/) (2026)
- [AFSS: Artifact-Focused Self-Synthesis for Mitigating Bias in Audio Deepfake Detection](https://mlsanity.com/paper/2603.26856/) (2026)
- [The Impact of Ideological Discourses in RAG: A Case Study with COVID-19 Treatments](https://mlsanity.com/paper/2603.14838/) (2026)

## Built on by (most cited first, 1 of 1)

- [What You Train Is What You Get: Gender Bias, Training Composition, and Post-Hoc Mitigation in Audio Deepfake Detection](https://mlsanity.com/paper/2607.09891/) (2026)

## Abstract (arXiv)

Audio deepfake detection systems are increasingly deployed in high-stakes security applications, yet their fairness across demographic groups remains critically underexamined. Prior work measures gender disparity but does not investigate where it comes from or how to fix it systematically. We present the first diagnosis-first framework that identifies bias source before applying targeted mitigation, evaluated on two models, AASIST and Wav2Vec2+ResNet18, on ASVSpoof5. Our diagnosis shows that bias does not stem from imbalanced training data but from acoustic representation differences, gender leakage in learned features, and structural evaluation asymmetry. We test mitigation strategies across in-processing, post-processing and combined families, including novel methods introduced in this work. Adjusting the decision threshold separately per gender reduces unfairness by 54% to 75% at no cost to detection accuracy, and our new epoch-level fairness regularisation method outperforms existing per-batch approaches. Adversarial debiasing succeeds only when gender leakage is localised, and fails when it is diffuse, an outcome correctly predicted by our diagnosis before training. No single method fully closes the fairness gap, confirming that bias sources must be identified before fixes are applied and that fairer benchmark design is equally important