# SARA: Stress Test Reasoning in Audio Deepfake Detection

arXiv: https://arxiv.org/abs/2601.03615 · HTML: https://mlsanity.com/paper/2601.03615/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Placed in Audio-language model reasoning, one of 17 directions in LLM reasoning.
- Builds on 8 of its 31 references that are in the map.
- First posted on arXiv 7 Jan 2026.

## Placed in

- LLM reasoning / [Audio-language model reasoning](https://mlsanity.com/lr/audio-language-model-reasoning/)

## Builds on (most cited first, 8 of 8)

- [Gemma 3 Technical Report](https://mlsanity.com/paper/2503.19786/) (2025)
- [Qwen2-Audio Technical Report](https://mlsanity.com/paper/2407.10759/) (2024)
- [Phi-4-Mini Technical Report: Compact yet Powerful Multimodal Language Models via Mixture-of-LoRAs](https://mlsanity.com/paper/2503.01743/) (2025)
- [MMAU-Pro: A Challenging and Comprehensive Benchmark for Holistic Evaluation of Audio General Intelligence](https://mlsanity.com/paper/2508.13992/) (2025)
- [SpeechR: A Benchmark for Speech Reasoning in Large Audio-Language Models](https://mlsanity.com/paper/2508.02018/) (2025)
- [Adversarial Attacks on Audio Deepfake Detection: A Benchmark and Comparative Study](https://mlsanity.com/paper/2509.07132/) (2025)
- [Toward Robust Real-World Audio Deepfake Detection: Closing the Explainability Gap](https://mlsanity.com/paper/2410.07436/) (2024)
- [FakeSound2: A Benchmark for Explainable and Generalizable Deepfake Sound Detection](https://mlsanity.com/paper/2509.17162/) (2025)

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

- [Towards Robust Speech Deepfake Detection via Human-Inspired Reasoning](https://mlsanity.com/paper/2603.10725/) (2026)
- [JECA^2: Judgment-Explanation Consistent Adversarial Attack against Forensic Vision-Language Models](https://mlsanity.com/paper/2605.28609/) (2026)

## Abstract (arXiv)

Audio Language Models (ALMs) offer a promising shift towards explainable audio deepfake detections (ADD), moving beyond textitblack-box classifiers by providing transparency to their predictions via reasoning traces. However, such reasoning may not support the model predictions, reflecting poor coherence, or, worse, may rationalize incorrect predictions with plausible but misleading explanation. Moreover, the behavior of ALM reasoning under adversarial attacks remains under-explored, raising questions about the practical reliability of such explanation capabilities. To address this gap, this study introduces SARA (Shift Analysis of Reasoning in Audio), a diagnostic framework that evaluates ALM reasoning across three dimensions: acoustic perception, reasoning-verdict coherence and dissonance. We test five open-source ALMs against both acoustic and linguistic adversarial attacks. We show that acoustic attacks significantly degrade reasoning-verdict coherence (average decrease of 14.20%), frequently inducing internal logical conflicts. Conversely, linguistic attacks achieve higher attack success rates while maintaining reasoning coherence. We further demonstrate that the textual coherence of generated reasoning traces also serves as a latent indicator of adversarial inputs, enabling effective detection of perturbed audio (0.78 in F1) textitwithout accessing the raw acoustic signal. These findings suggest that reasoning traces provide diagnostic utility that persists even when final classification outputs are compromised.