arXiv:2606. 30780v1 Announce Type: cross Abstract: Audio deepfakes are a growing challenge for the general public, as well as for journalists and fact-checkers.
arXiv:2610.16049v1 Announce Type: cross Abstract: Existing audio deepfake detection (ADD) methods frequently exhibit limited generalization, as fixed bottleneck architectures fail to adapt to the het...
arXiv:2603. 10725v3 Announce Type: replace-cross Abstract: The modern generative audio models can be used by an adversary in an unlawful manner, specifically, to impersonate other people to gain access to private information.
ToolDF is a tool‑integrated reasoning framework designed for detecting mixed‑authenticity audio deepfakes, where genuine and manipulated audio cues coexist across time or overlapping sources. It uses an audio large language model to orchestrate tasks such as source separation and routing to domain‑specific experts, aggregating their evidence into an interpretable verdict. The authors also introduce a mixed‑authenticity ADD benchmark and report that ToolDF outperforms monolithic baselines, achieving significant macro‑F1 gains while localizing evidence to specific temporal regions and acoustic sources.
Samsone is a family of small audio language models (SALMs) designed for on‑device inference, with the flagship Samsone‑134M setting a new state‑of‑the‑art for its size class across multiple benchmarks. The paper also presents Samsone‑99M and Samsone‑356M to study scaling laws, showing that these compact models achieve performance competitive with much larger counterparts. The authors train the models on publicly available data and release training code, weights, mobile‑optimized checkpoints, and an open‑source Android app for real‑time inference.
arXiv:2606. 10246v1 Announce Type: cross Abstract: Maliciously-created fake speech, including deepfaked and spoofed audio, is proliferating at an alarming rate, and detection models are racing to stay ahead of the curve.
arXiv:2606. 29544v1 Announce Type: cross Abstract: We present Proteus, a framework developed at Resemble AI for automated robustness testing of our audio deepfake detection system.
The paper investigates how to improve self‑supervised learning (SSL) for Finnish speech by filtering broadcast audio based on automatic speech recognition (ASR) confidence rather than traditional voice activity detection or language identification. Using 2.1 million hours of raw broadcast audio, the authors demonstrate that ASR‑confidence filtering yields 3–7 % relative WER gains over LID filtering while training on a smaller dataset. They also conduct a controlled comparison of four SSL architectures—Transformer, Macaron‑Net, Conformer, and E‑Branchformer—at matched parameter budgets, showing that their models surpass previous Finnish SSL baselines, especially at Base scale and on out‑of‑domain data.
arXiv:2606. 10912v1 Announce Type: cross Abstract: 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.
The paper proposes a framework for detecting deepfake audio in voice authentication systems at the network edge. It combines interpretable physics-based features that model vocal tract dynamics with representations from a self‑supervised learning module, processed through a streamlined MLP and a Bayesian ensemble to provide uncertainty estimates. The approach also includes a trust‑based aggregation protocol to protect the control plane from poisoning attacks in federated learning environments.
CoRELoop introduces a parameter‑efficient refinement framework for audio deepfake detection that operates on a pre‑trained SSL‑based detector without altering its original parameters. By adapting recurrent inputs, controlling state updates, and aligning refined outputs with the frozen classifier, CoRELoop adds lightweight refinement modules and low‑rank adapters, achieving a pooled equal error rate reduction from 4.85% to 3.74% on 14 cross‑domain test sets with only about 10 M trainable parameters. An optional halting head further optimizes performance, reaching 3.73% pooled EER with an average of 1.18 passes.
arXiv:2607. 28351v2 Announce Type: replace-cross Abstract: Speech deepfake detection has expanded in scope with increasingly heterogeneous spoofing mechanisms, including speech synthesis, voice conversion, vocoder reconstruction, and neural-codec resynthesis.
arXiv:2607. 03418v1 Announce Type: cross Abstract: A trustworthy and GDPR-compliant deepfake audio detector must base its decisions on acoustic artifacts, not on what is being said or who is speaking.