# What Do Deepfake Speech Detectors Actually Hear?

arXiv: https://arxiv.org/abs/2606.10912 · HTML: https://mlsanity.com/paper/2606.10912/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 19 of its 31 references that are in the map.
- First posted on arXiv 9 Jun 2026.

## Builds on (most cited first, 19 of 19)

- ["Why Should I Trust You?": Explaining the Predictions of Any Classifier](https://mlsanity.com/paper/1602.04938/) (2016)
- [wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations](https://mlsanity.com/paper/2006.11477/) (2020)
- [Axiomatic Attribution for Deep Networks](https://mlsanity.com/paper/1703.01365/) (2017)
- [WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech Processing](https://mlsanity.com/paper/2110.13900/) (2021)
- [YourTTS: Towards Zero-Shot Multi-Speaker TTS and Zero-Shot Voice Conversion for everyone](https://mlsanity.com/paper/2112.02418/) (2021)
- [AASIST: Audio Anti-Spoofing using Integrated Spectro-Temporal Graph Attention Networks](https://mlsanity.com/paper/2110.01200/) (2021)
- [ASVspoof 5: Crowdsourced Speech Data, Deepfakes, and Adversarial Attacks at Scale](https://mlsanity.com/paper/2408.08739/) (2024)
- [RawBoost: A Raw Data Boosting and Augmentation Method applied to Automatic Speaker Verification Anti-Spoofing](https://mlsanity.com/paper/2111.04433/) (2021)
- [Explaining deep learning models for spoofing and deepfake detection with SHapley Additive exPlanations](https://mlsanity.com/paper/2110.03309/) (2021)
- [BUT Systems and Analyses for the ASVspoof 5 Challenge](https://mlsanity.com/paper/2408.11152/) (2024)
- [Exploring WavLM Back-ends for Speech Spoofing and Deepfake Detection](https://mlsanity.com/paper/2409.05032/) (2024)
- [AASIST3: KAN-Enhanced AASIST Speech Deepfake Detection using SSL Features and Additional Regularization for the ASVspoof 2024 Challenge](https://mlsanity.com/paper/2408.17352/) (2024)
- [Diffuse or Confuse: A Diffusion Deepfake Speech Dataset](https://mlsanity.com/paper/2410.06796/) (2024)
- [SCDF: A Speaker Characteristics DeepFake Speech Dataset for Bias Analysis](https://mlsanity.com/paper/2508.07944/) (2025)
- [Explainable deepfake and spoofing detection: an attack analysis using SHapley Additive exPlanations](https://mlsanity.com/paper/2202.13693/) (2022)
- [SZU-AFS Antispoofing System for the ASVspoof 5 Challenge](https://mlsanity.com/paper/2408.09933/) (2024)
- [CA-MHFA: A Context-Aware Multi-Head Factorized Attentive Pooling for SSL-Based Speaker Verification](https://mlsanity.com/paper/2409.15234/) (2024)
- [Evolutionary Multi-Objective Fusion of Deepfake Speech Detectors](https://mlsanity.com/paper/2604.01330/) (2026)
- [Hybrid Pruning: In-Situ Compression of Self-Supervised Speech Models for Speaker Verification and Anti-Spoofing](https://mlsanity.com/paper/2508.16232/) (2025)

## Abstract (arXiv)

Deepfake speech detectors often output a single score without explaining why an audio sample is flagged, where in the signal the evidence lies, or what cues drive the decision. We propose an audio-native explainability pipeline using Integrated Gradients on time-aligned self-supervised representations to localize decision evidence over time. We apply the proposed method to three WavLM-based detectors (AASIST, CA-MHFA, SLS) on ASVspoof 5 and manually annotate the highest-attribution regions to provide a semantic meaning of the most important cues. Despite similar performance, the detectors rely on different cues: AASIST emphasizes non-speech/environment cues, CA-MHFA focuses on localized phoneme artifacts, and SLS relies on word boundaries and spectral integrity. We move beyond speculative reasoning and validate our findings by causal masking of the primary detector cues. Observed performance degradation further supports the explained detector semantics.