arXiv subjects

Paolo Bestagini

Publications and source records attributed to Paolo Bestagini.

At least 19 recordsLinked to original sources

arXiv2026

Toward Interpretable Speech Deepfake Detection using Artifact-Specific Experts and Calibrated Detection Scores

In this work, we propose an interpretable framework for speech deepfake detection based on artifact-specific expert models. Rather than relying on black-box decisions, the framework provides human-understandable evidence, which is critical in high-stakes settings. Each expert is trained to detect a specific speech synthesis artifact, and its output is calibrated into a log-likelihood ratio that serves as an interpretable evidence score. We evaluate five artifact-specific experts and show that, with proper calibration, they can capture their target artifacts and produce meaningful evidence. Importantly, each expert estimates only the presence of its assigned artifact rather than directly performing the final decision. Their outputs are aggregated into an ensemble to produce the actual real-versus-fake classification, while maintaining interpretability by indicating how strongly each expert supports or contradicts a fake classification. Results show that artifact-specific experts capture interpretable signals of synthetic speech across multiple generation pipelines.

cs.SD↗

arXiv2026

CUPID: Reconstructing UV Texture Maps for Interpretable Person-of-Interest Deepfake Detection

Deepfakes targeting a high-profile individual, known as Person-of-Interest (POI), are a threat to modern democracies and societies. Current POI deepfake detection methods still struggle to combine robustness to post-processing, efficiency and interpretability, key aspects of modern deepfake detectors. In this paper we propose CUPID, a POI video deepfake detector that combines UV texture maps, a facial appearance representation derived from 3D face reconstructions, with the representation learning capabilities of the Masked Autoencoder (MAE). Our method does not require any deepfake videos in its training phase. Moreover, it does not even require including a specific POI in the training set: the combination of UV texture maps extracted from real video frames and the MAE context-guided reconstruction yields a latent space that captures rich and discriminative facial features even for identities unseen during training. In the testing phase, the embeddings extracted from a query video depicting the POI can be matched against pristine reference videos to assess the video authenticity. Furthermore, operating in the UV space naturally provides an additional layer of interpretability. Specifically, we can extract decoded residual maps that highlight which facial regions of a test video deviate most from the identity representation of the corresponding POI. Experiments on four deepfake datasets show that CUPID outperforms the current state of the art on most datasets and achieves the best overall robustness against strong downscaling and compression, while also providing substantially faster inference. Our experimental code will be released at https://github.com/polimi-ispl/CUPID.

cs.CV↗

arXiv2026

Multi-Task Transformer for Explainable Speech Deepfake Detection via Formant Modeling

In this work, we introduce a multi-task transformer for speech deepfake detection, capable of predicting formant trajectories and voicing patterns over time, ultimately classifying speech as real or fake, and highlighting whether its decisions rely more on voiced or unvoiced regions. Building on a prior speaker-formant transformer architecture, we streamline the model with an improved input segmentation strategy, redesign the decoding process, and integrate built-in explainability. Compared to the baseline, our model requires fewer parameters, trains faster, and provides better interpretability, without sacrificing prediction performance.

cs.SD↗

arXiv2025

Enhanced Water Leak Detection with Convolutional Neural Networks and One-Class Support Vector Machine

Water is a critical resource that must be managed efficiently. However, a substantial amount of water is lost each year due to leaks in Water Distribution Networks (WDNs). This underscores the need for reliable and effective leak detection and localization systems. In recent years, various solutions have been proposed, with data-driven approaches gaining increasing attention due to their superior performance. In this paper, we propose a new method for leak detection. The method is based on water pressure measurements acquired at a series of nodes of a WDN. Our technique is a fully data-driven solution that makes only use of the knowledge of the WDN topology, and a series of pressure data acquisitions obtained in absence of leaks. The proposed solution is based on an feature extractor and a one-class Support Vector Machines (SVM) trained on no-leak data, so that leaks are detected as anomalies. The results achieved on a simulate dataset using the Modena WDN demonstrate that the proposed solution outperforms recent methods for leak detection.

cs.LG↗

arXiv2025

Beyond Spectral Peaks: Interpreting the Cues Behind Synthetic Image Detection

Over the years, the forensics community has proposed several deep learning-based detectors to mitigate the risks of generative AI. Recently, frequency-domain artifacts (particularly periodic peaks in the magnitude spectrum), have received significant attention, as they have been often considered a strong indicator of synthetic image generation. However, state-of-the-art detectors are typically used as black-boxes, and it still remains unclear whether they truly rely on these peaks. This limits their interpretability and trust. In this work, we conduct a systematic study to address this question. We propose a strategy to remove spectral peaks from images and analyze the impact of this operation on several detectors. In addition, we introduce a simple linear detector that relies exclusively on frequency peaks, providing a fully interpretable baseline free from the confounding influence of deep learning. Our findings reveal that most detectors are not fundamentally dependent on spectral peaks, challenging a widespread assumption in the field and paving the way for more transparent and reliable forensic tools.

cs.CV↗

arXiv2025

Forensic Similarity for Speech Deepfakes

In this paper, we introduce the concept of forensic similarity in the speech deepfake detection domain, which aims to determine whether two audio segments share the same underlying forensic traces. Our approach is inspired by prior work in the image domain. To transfer this idea to the audio domain, we propose a two-stage deep learning framework consisting of a Siamese-based feature extractor and a core decision module, referred to as the similarity network. The system goal to assess whether two speech samples originate from the same source by comparing their forensic characteristics. In practice, the model maps pairs of audio segments to a similarity score indicating whether they contain identical or different forensic traces. We evaluate the proposed method on the emerging task of source verification, demonstrating its ability to determine whether two speech samples were generated by the same model. In addition, we explore its applicability to audio splicing detection as a complementary use case. Experimental results show that the proposed approach generalizes well to previously unseen forensic traces, highlighting its robustness, flexibility, and practical relevance for digital audio forensics.

cs.SD↗

arXiv2025

Attention-based Mixture of Experts for Robust Speech Deepfake Detection

AI-generated speech is becoming increasingly used in everyday life, powering virtual assistants, accessibility tools, and other applications. However, it is also being exploited for malicious purposes such as impersonation, misinformation, and biometric spoofing. As speech deepfakes become nearly indistinguishable from real human speech, the need for robust detection methods and effective countermeasures has become critically urgent. In this paper, we present the ISPL's submission to the SAFE challenge at IH&MMSec 2025, where our system ranked first across all tasks. Our solution introduces a novel approach to audio deepfake detection based on a Mixture of Experts architecture. The proposed system leverages multiple state-of-the-art detectors, combining their outputs through an attention-based gating network that dynamically weights each expert based on the input speech signal. In this design, each expert develops a specialized understanding of the shared training data by learning to capture different complementary aspects of the same input through inductive biases. Experimental results indicate that our method outperforms existing approaches across multiple datasets. We further evaluate and analyze the performance of our system in the SAFE challenge.

cs.SD↗

arXiv2025

Mitigating data replication in text-to-audio generative diffusion models through anti-memorization guidance

A persistent challenge in generative audio models is data replication, where the model unintentionally generates parts of its training data during inference. In this work, we address this issue in text-to-audio diffusion models by exploring the use of anti-memorization strategies. We adopt Anti-Memorization Guidance (AMG), a technique that modifies the sampling process of pre-trained diffusion models to discourage memorization. Our study explores three types of guidance within AMG, each designed to reduce replication while preserving generation quality. We use Stable Audio Open as our backbone, leveraging its fully open-source architecture and training dataset. Our comprehensive experimental analysis suggests that AMG significantly mitigates memorization in diffusion-based text-to-audio generation without compromising audio fidelity or semantic alignment.

eess.AS↗

arXiv2025

Phoneme-Level Analysis for Person-of-Interest Speech Deepfake Detection

Recent advances in generative AI have made the creation of speech deepfakes widely accessible, posing serious challenges to digital trust. To counter this, various speech deepfake detection strategies have been proposed, including Person-of-Interest (POI) approaches, which focus on identifying impersonations of specific individuals by modeling and analyzing their unique vocal traits. Despite their excellent performance, the existing methods offer limited granularity and lack interpretability. In this work, we propose a POI-based speech deepfake detection method that operates at the phoneme level. Our approach decomposes reference audio into phonemes to construct a detailed speaker profile. In inference, phonemes from a test sample are individually compared against this profile, enabling fine-grained detection of synthetic artifacts. The proposed method achieves comparable accuracy to traditional approaches while offering superior robustness and interpretability, key aspects in multimedia forensics. By focusing on phoneme analysis, this work explores a novel direction for explainable, speaker-centric deepfake detection.

cs.SD↗

arXiv2025

Source Verification for Speech Deepfakes

With the proliferation of speech deepfake generators, it becomes crucial not only to assess the authenticity of synthetic audio but also to trace its origin. While source attribution models attempt to address this challenge, they often struggle in open-set conditions against unseen generators. In this paper, we introduce the source verification task, which, inspired by speaker verification, determines whether a test track was produced using the same model as a set of reference signals. Our approach leverages embeddings from a classifier trained for source attribution, computing distance scores between tracks to assess whether they originate from the same source. We evaluate multiple models across diverse scenarios, analyzing the impact of speaker diversity, language mismatch, and post-processing operations. This work provides the first exploration of source verification, highlighting its potential and vulnerabilities, and offers insights for real-world forensic applications.

cs.SD↗

arXiv2025

WILD: a new in-the-Wild Image Linkage Dataset for synthetic image attribution

Synthetic image source attribution is an open challenge, with an increasing number of image generators being released yearly. The complexity and the sheer number of available generative techniques, as well as the scarcity of high-quality open source datasets of diverse nature for this task, make training and benchmarking synthetic image source attribution models very challenging. WILD is a new in-the-Wild Image Linkage Dataset designed to provide a powerful training and benchmarking tool for synthetic image attribution models. The dataset is built out of a closed set of 10 popular commercial generators, which constitutes the training base of attribution models, and an open set of 10 additional generators, simulating a real-world in-the-wild scenario. Each generator is represented by 1,000 images, for a total of 10,000 images in the closed set and 10,000 images in the open set. Half of the images are post-processed with a wide range of operators. WILD allows benchmarking attribution models in a wide range of tasks, including closed and open set identification and verification, and robust attribution with respect to post-processing and adversarial attacks. Models trained on WILD are expected to benefit from the challenging scenario represented by the dataset itself. Moreover, an assessment of seven baseline methodologies on closed and open set attribution is presented, including robustness tests with respect to post-processing.

cs.MM↗

arXiv2025

Leveraging Land Cover Priors for Isoprene Emission Super-Resolution

Remote sensing plays a crucial role in monitoring Earth's ecosystems, yet satellite-derived data often suffer from limited spatial resolution, restricting their applicability in atmospheric modeling and climate research. In this work, we propose a deep learning-based Super-Resolution (SR) framework that leverages land cover information to enhance the spatial accuracy of Biogenic Volatile Organic Compounds (BVOCs) emissions, with a particular focus on isoprene. Our approach integrates land cover priors as emission drivers, capturing spatial patterns more effectively than traditional methods. We evaluate the model's performance across various climate conditions and analyze statistical correlations between isoprene emissions and key environmental information such as cropland and tree cover data. Additionally, we assess the generalization capabilities of our SR model by applying it to unseen climate zones and geographical regions. Experimental results demonstrate that incorporating land cover data significantly improves emission SR accuracy, particularly in heterogeneous landscapes. This study contributes to atmospheric chemistry and climate modeling by providing a cost-effective, data-driven approach to refining BVOC emission maps. The proposed method enhances the usability of satellite-based emissions data, supporting applications in air quality forecasting, climate impact assessments, and environmental studies.

cs.CV↗

arXiv2025

Anomaly Detection and Localization for Speech Deepfakes via Feature Pyramid Matching

The rise of AI-driven generative models has enabled the creation of highly realistic speech deepfakes - synthetic audio signals that can imitate target speakers' voices - raising critical security concerns. Existing methods for detecting speech deepfakes primarily rely on supervised learning, which suffers from two critical limitations: limited generalization to unseen synthesis techniques and a lack of explainability. In this paper, we address these issues by introducing a novel interpretable one-class detection framework, which reframes speech deepfake detection as an anomaly detection task. Our model is trained exclusively on real speech to characterize its distribution, enabling the classification of out-of-distribution samples as synthetically generated. Additionally, our framework produces interpretable anomaly maps during inference, highlighting anomalous regions across both time and frequency domains. This is done through a Student-Teacher Feature Pyramid Matching system, enhanced with Discrepancy Scaling to improve generalization capabilities across unseen data distributions. Extensive evaluations demonstrate the superior performance of our approach compared to the considered baselines, validating the effectiveness of framing speech deepfake detection as an anomaly detection problem.

cs.SD↗

arXiv2025

Is JPEG AI going to change image forensics?

In this paper, we investigate the counter-forensic effects of the new JPEG AI standard based on neural image compression, focusing on two critical areas: deepfake image detection and image splicing localization. Neural image compression leverages advanced neural network algorithms to achieve higher compression rates while maintaining image quality. However, it introduces artifacts that closely resemble those generated by image synthesis techniques and image splicing pipelines, complicating the work of researchers when discriminating pristine from manipulated content. We comprehensively analyze JPEG AI's counter-forensic effects through extensive experiments on several state-of-the-art detectors and datasets. Our results demonstrate a reduction in the performance of leading forensic detectors when analyzing content processed through JPEG AI. By exposing the vulnerabilities of the available forensic tools, we aim to raise the urgent need for multimedia forensics researchers to include JPEG AI images in their experimental setups and develop robust forensic techniques to distinguish between neural compression artifacts and actual manipulations.

eess.IV↗

arXiv2025

Hiding Local Manipulations on SAR Images: a Counter-Forensic Attack

The vast accessibility of Synthetic Aperture Radar (SAR) images through online portals has propelled the research across various fields. This widespread use and easy availability have unfortunately made SAR data susceptible to malicious alterations, such as local editing applied to the images for inserting or covering the presence of sensitive targets. Vulnerability is further emphasized by the fact that most SAR products, despite their original complex nature, are often released as amplitude-only information, allowing even inexperienced attackers to edit and easily alter the pixel content. To contrast malicious manipulations, in the last years the forensic community has begun to dig into the SAR manipulation issue, proposing detectors that effectively localize the tampering traces in amplitude images. Nonetheless, in this paper we demonstrate that an expert practitioner can exploit the complex nature of SAR data to obscure any signs of manipulation within a locally altered amplitude image. We refer to this approach as a counter-forensic attack. To achieve the concealment of manipulation traces, the attacker can simulate a re-acquisition of the manipulated scene by the SAR system that initially generated the pristine image. In doing so, the attacker can obscure any evidence of manipulation, making it appear as if the image was legitimately produced by the system. This attack has unique features that make it both highly generalizable and relatively easy to apply. First, it is a black-box attack, meaning it is not designed to deceive a specific forensic detector. Furthermore, it does not require a training phase and is not based on adversarial operations. We assess the effectiveness of the proposed counter-forensic approach across diverse scenarios, examining various manipulation operations.

cs.CV↗

arXiv2024

Comparative Analysis of ASR Methods for Speech Deepfake Detection

Recent techniques for speech deepfake detection often rely on pre-trained self-supervised models. These systems, initially developed for Automatic Speech Recognition (ASR), have proved their ability to offer a meaningful representation of speech signals, which can benefit various tasks, including deepfake detection. In this context, pre-trained models serve as feature extractors and are used to extract embeddings from input speech, which are then fed to a binary speech deepfake detector. The remarkable accuracy achieved through this approach underscores a potential relationship between ASR and speech deepfake detection. However, this connection is not yet entirely clear, and we do not know whether improved performance in ASR corresponds to higher speech deepfake detection capabilities. In this paper, we address this question through a systematic analysis. We consider two different pre-trained self-supervised ASR models, Whisper and Wav2Vec 2.0, and adapt them for the speech deepfake detection task. These models have been released in multiple versions, with increasing number of parameters and enhanced ASR performance. We investigate whether performance improvements in ASR correlate with improvements in speech deepfake detection. Our results provide insights into the relationship between these two tasks and offer valuable guidance for the development of more effective speech deepfake detectors.

cs.SD↗

arXiv2024

POLIPHONE: A Dataset for Smartphone Model Identification from Audio Recordings

When dealing with multimedia data, source attribution is a key challenge from a forensic perspective. This task aims to determine how a given content was captured, providing valuable insights for various applications, including legal proceedings and integrity investigations. The source attribution problem has been addressed in different domains, from identifying the camera model used to capture specific photographs to detecting the synthetic speech generator or microphone model used to create or record given audio tracks. Recent advancements in this area rely heavily on machine learning and data-driven techniques, which often outperform traditional signal processing-based methods. However, a drawback of these systems is their need for large volumes of training data, which must reflect the latest technological trends to produce accurate and reliable predictions. This presents a significant challenge, as the rapid pace of technological progress makes it difficult to maintain datasets that are up-to-date with real-world conditions. For instance, in the task of smartphone model identification from audio recordings, the available datasets are often outdated or acquired inconsistently, making it difficult to develop solutions that are valid beyond a research environment. In this paper we present POLIPHONE, a dataset for smartphone model identification from audio recordings. It includes data from 20 recent smartphones recorded in a controlled environment to ensure reproducibility and scalability for future research. The released tracks contain audio data from various domains (i.e., speech, music, environmental sounds), making the corpus versatile and applicable to a wide range of use cases. We also present numerous experiments to benchmark the proposed dataset using a state-of-the-art classifier for smartphone model identification from audio recordings.

cs.SD↗

arXiv2024

Explainable Artifacts for Synthetic Western Blot Source Attribution

Recent advancements in artificial intelligence have enabled generative models to produce synthetic scientific images that are indistinguishable from pristine ones, posing a challenge even for expert scientists habituated to working with such content. When exploited by organizations known as paper mills, which systematically generate fraudulent articles, these technologies can significantly contribute to the spread of misinformation about ungrounded science, potentially undermining trust in scientific research. While previous studies have explored black-box solutions, such as Convolutional Neural Networks, for identifying synthetic content, only some have addressed the challenge of generalizing across different models and providing insight into the artifacts in synthetic images that inform the detection process. This study aims to identify explainable artifacts generated by state-of-the-art generative models (e.g., Generative Adversarial Networks and Diffusion Models) and leverage them for open-set identification and source attribution (i.e., pointing to the model that created the image).

cs.CV↗