arXiv subjects

Wanying Ge

Publications and source records attributed to Wanying Ge.

At least 19 recordsLinked to original sources

arXiv2026

A Training-Free Proactive Defense Against Partial Speech Manipulation via Self-Embedding Steganography

Partial deepfake speech, where only limited segments of an utterance are synthesized or manipulated, poses a significant challenge to existing deepfake detection systems. As the proportion of spoofed regions decreases, passive detectors become increasingly unreliable, and accurate detection and restoration remain challenging. In this paper, we revisit audio steganography from a new perspective and propose its use as a proactive defense against partially deepfaked audio. In particular, we consider a self-embedding strategy in which a clean speech signal embeds a compressed representation of itself, enabling post-hoc extraction of reference content. We demonstrate how existing audio steganography methods can be repurposed to support detection of partial deepfakes through codec-based restoration. Experiments on a benchmark dataset show that the proposed approach complements passive defenses. Remarkably, the proposed method operates without any training, providing a robust and data-efficient alternative for partial deepfake detection.

cs.SD↗

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

Deepfake Word Detection by Next-token Prediction using Fine-tuned Whisper

Deepfake speech utterances can be forged by replacing one or more words in a bona fide utterance with semantically different words synthesized with speech-generative models. While a dedicated synthetic word detector could be developed, we developed a cost-effective method that fine-tunes a pre-trained Whisper model to detect synthetic words while transcribing the input utterance via next-token prediction. We further investigate using partially vocoded utterances as the fine-tuning data, thus reducing the cost of data collection. Our experiments demonstrate that, on in-domain test data, the fine-tuned Whisper yields low synthetic-word detection error rates and transcription error rates. On out-of-domain test data with synthetic words produced with unseen speech-generative models, the fine-tuned Whisper remains on par with a dedicated ResNet-based detection model; however, the overall performance degradation calls for strategies to improve its generalization capability.

eess.AS↗

arXiv2026

Self Voice Conversion as an Attack against Neural Audio Watermarking

Audio watermarking embeds auxiliary information into speech while maintaining speaker identity, linguistic content, and perceptual quality. Although recent advances in neural and digital signal processing-based watermarking methods have improved imperceptibility and embedding capacity, robustness is still primarily assessed against conventional distortions such as compression, additive noise, and resampling. However, the rise of deep learning-based attacks introduces novel and significant threats to watermark security. In this work, we investigate self voice conversion as a universal, content-preserving attack against audio watermarking systems. Self voice conversion remaps a speaker's voice to the same identity while altering acoustic characteristics through a voice conversion model. We demonstrate that this attack severely degrades the reliability of state-of-the-art watermarking approaches and highlight its implications for the security of modern audio watermarking techniques.

cs.SD↗

arXiv2026

Summary of The Inaugural Music Source Restoration Challenge

Music Source Restoration (MSR) aims to recover original, unprocessed instrument stems from professionally mixed and degraded audio, requiring the reversal of both production effects and real-world degradations. We present the inaugural MSR Challenge, which features objective evaluation on studio-produced mixtures using Multi-Mel-SNR, Zimtohrli, and FAD-CLAP, alongside subjective evaluation on real-world degraded recordings. Five teams participated in the challenge. The winning system achieved 4.46 dB Multi-Mel-SNR and 3.47 MOS-Overall, corresponding to relative improvements of 91% and 18% over the second-place system, respectively. Per-stem analysis reveals substantial variation in restoration difficulty across instruments, with bass averaging 4.59 dB across all teams, while percussion averages only 0.29 dB. The dataset, evaluation protocols, and baselines are available at https://msrchallenge.com/.

cs.SD↗

arXiv2025

FakeMark: Deepfake Speech Attribution With Watermarked Artifacts

Deepfake speech attribution remains challenging for existing solutions. Classifier-based solutions often fail to generalize to domain-shifted samples, and watermarking-based solutions are easily compromised by distortions like codec compression or malicious removal attacks. To address these issues, we propose FakeMark, a novel watermarking framework that injects artifact-correlated watermarks associated with deepfake systems rather than pre-assigned bitstring messages. This design allows a detector to attribute the source system by leveraging both injected watermark and intrinsic deepfake artifacts, remaining effective even if one of these cues is elusive or removed. Experimental results show that FakeMark improves generalization to cross-dataset samples where classifier-based solutions struggle and maintains high accuracy under various distortions where conventional watermarking-based solutions fail.

eess.AS↗

arXiv2025

MSRBench: A Benchmarking Dataset for Music Source Restoration

Music Source Restoration (MSR) extends source separation to realistic settings where signals undergo production effects (equalization, compression, reverb) and real-world degradations, with the goal of recovering the original unprocessed sources. Existing benchmarks cannot measure restoration fidelity: synthetic datasets use unprocessed stems but unrealistic mixtures, while real production datasets provide only already-processed stems without clean references. We present MSRBench, the first benchmark explicitly designed for MSR evaluation. MSRBench contains raw stem-mixture pairs across eight instrument classes, where mixtures are produced by professional mixing engineers. These raw-processed pairs enable direct evaluation of both separation accuracy and restoration fidelity. Beyond controlled studio conditions, the mixtures are augmented with twelve real-world degradations spanning analog artifacts, acoustic environments, and lossy codecs. Baseline experiments with U-Net and BSRNN achieve SI-SNR of -37.8 dB and -23.4 dB respectively, with perceptual quality (FAD CLAP) around 0.7-0.8, demonstrating substantial room for improvement and the need for restoration-specific architectures.

cs.SD↗

arXiv2025

Towards Data Drift Monitoring for Speech Deepfake Detection in the context of MLOps

When being delivered in applications or services on the cloud, static speech deepfake detectors that are not updated will become vulnerable to newly created speech deepfake attacks. From the perspective of machine learning operations (MLOps), this paper tries to answer whether we can monitor new and unseen speech deepfake data that drifts away from a seen reference data set. We further ask, if drift is detected, whether we can fine-tune the detector using similarly drifted data, reduce the drift, and improve the detection performance. On a toy dataset and the large-scale MLAAD dataset, we show that the drift caused by new text-to-speech (TTS) attacks can be monitored using distances between the distributions of the new data and reference data. Furthermore, we demonstrate that fine-tuning the detector using data generated by the new TTS deepfakes can reduce the drift and the detection error rates.

eess.AS↗

arXiv2025

LENS-DF: Deepfake Detection and Temporal Localization for Long-Form Noisy Speech

This study introduces LENS-DF, a novel and comprehensive recipe for training and evaluating audio deepfake detection and temporal localization under complicated and realistic audio conditions. The generation part of the recipe outputs audios from the input dataset with several critical characteristics, such as longer duration, noisy conditions, and containing multiple speakers, in a controllable fashion. The corresponding detection and localization protocol uses models. We conduct experiments based on self-supervised learning front-end and simple back-end. The results indicate that models trained using data generated with LENS-DF consistently outperform those trained via conventional recipes, demonstrating the effectiveness and usefulness of LENS-DF for robust audio deepfake detection and localization. We also conduct ablation studies on the variations introduced, investigating their impact on and relevance to realistic challenges in the field.

cs.SD↗

arXiv2025

Post-training for Deepfake Speech Detection

We introduce a post-training approach that adapts self-supervised learning (SSL) models for deepfake speech detection by bridging the gap between general pre-training and domain-specific fine-tuning. We present AntiDeepfake models, a series of post-trained models developed using a large-scale multilingual speech dataset containing over 56,000 hours of genuine speech and 18,000 hours of speech with various artifacts in over one hundred languages. Experimental results show that the post-trained models already exhibit strong robustness and generalization to unseen deepfake speech. When they are further fine-tuned on the Deepfake-Eval-2024 dataset, these models consistently surpass existing state-of-the-art detectors that do not leverage post-training. Model checkpoints and source code are available online.

eess.AS↗

arXiv2025

A Comparative Study on Proactive and Passive Detection of Deepfake Speech

Solutions for defending against deepfake speech fall into two categories: proactive watermarking models and passive conventional deepfake detectors. While both address common threats, their differences in training, optimization, and evaluation prevent a unified protocol for joint evaluation and selecting the best solutions for different cases. This work proposes a framework to evaluate both model types in deepfake speech detection. To ensure fair comparison and minimize discrepancies, all models were trained and tested on common datasets, with performance evaluated using a shared metric. We also analyze their robustness against various adversarial attacks, showing that different models exhibit distinct vulnerabilities to different speech attribute distortions. Our training and evaluation code is available at Github.

cs.SD↗

arXiv2023

Spoofing attack augmentation: can differently-trained attack models improve generalisation?

A reliable deepfake detector or spoofing countermeasure (CM) should be robust in the face of unpredictable spoofing attacks. To encourage the learning of more generaliseable artefacts, rather than those specific only to known attacks, CMs are usually exposed to a broad variety of different attacks during training. Even so, the performance of deep-learning-based CM solutions are known to vary, sometimes substantially, when they are retrained with different initialisations, hyper-parameters or training data partitions. We show in this paper that the potency of spoofing attacks, also deep-learning-based, can similarly vary according to training conditions, sometimes resulting in substantial degradations to detection performance. Nevertheless, while a RawNet2 CM model is vulnerable when only modest adjustments are made to the attack algorithm, those based upon graph attention networks and self-supervised learning are reassuringly robust. The focus upon training data generated with different attack algorithms might not be sufficient on its own to ensure generaliability; some form of spoofing attack augmentation at the algorithm level can be complementary.

cs.CR↗

arXiv2023

Malafide: a novel adversarial convolutive noise attack against deepfake and spoofing detection systems

We present Malafide, a universal adversarial attack against automatic speaker verification (ASV) spoofing countermeasures (CMs). By introducing convolutional noise using an optimised linear time-invariant filter, Malafide attacks can be used to compromise CM reliability while preserving other speech attributes such as quality and the speaker's voice. In contrast to other adversarial attacks proposed recently, Malafide filters are optimised independently of the input utterance and duration, are tuned instead to the underlying spoofing attack, and require the optimisation of only a small number of filter coefficients. Even so, they degrade CM performance estimates by an order of magnitude, even in black-box settings, and can also be configured to overcome integrated CM and ASV subsystems. Integrated solutions that use self-supervised learning CMs, however, are more robust, under both black-box and white-box settings.

eess.AS↗

arXiv2023

Can spoofing countermeasure and speaker verification systems be jointly optimised?

Spoofing countermeasure (CM) and automatic speaker verification (ASV) sub-systems can be used in tandem with a backend classifier as a solution to the spoofing aware speaker verification (SASV) task. The two sub-systems are typically trained independently to solve different tasks. While our previous work demonstrated the potential of joint optimisation, it also showed a tendency to over-fit to speakers and a lack of sub-system complementarity. Using only a modest quantity of auxiliary data collected from new speakers, we show that joint optimisation degrades the performance of separate CM and ASV sub-systems, but that it nonetheless improves complementarity, thereby delivering superior SASV performance. Using standard SASV evaluation data and protocols, joint optimisation reduces the equal error rate by 27\% relative to performance obtained using fixed, independently-optimised sub-systems under like-for-like training conditions.

eess.AS↗

arXiv2022

On the potential of jointly-optimised solutions to spoofing attack detection and automatic speaker verification

The spoofing-aware speaker verification (SASV) challenge was designed to promote the study of jointly-optimised solutions to accomplish the traditionally separately-optimised tasks of spoofing detection and speaker verification. Jointly-optimised systems have the potential to operate in synergy as a better performing solution to the single task of reliable speaker verification. However, none of the 23 submissions to SASV 2022 are jointly optimised. We have hence sought to determine why separately-optimised sub-systems perform best or why joint optimisation was not successful. Experiments reported in this paper show that joint optimisation is successful in improving robustness to spoofing but that it degrades speaker verification performance. The findings suggest that spoofing detection and speaker verification sub-systems should be optimised jointly in a manner which reflects the differences in how information provided by each sub-system is complementary to that provided by the other. Progress will also likely depend upon the collection of data from a larger number of speakers.

eess.AS↗

arXiv2022

Explainable deepfake and spoofing detection: an attack analysis using SHapley Additive exPlanations

Despite several years of research in deepfake and spoofing detection for automatic speaker verification, little is known about the artefacts that classifiers use to distinguish between bona fide and spoofed utterances. An understanding of these is crucial to the design of trustworthy, explainable solutions. In this paper we report an extension of our previous work to better understand classifier behaviour to the use of SHapley Additive exPlanations (SHAP) to attack analysis. Our goal is to identify the artefacts that characterise utterances generated by different attacks algorithms. Using a pair of classifiers which operate either upon raw waveforms or magnitude spectrograms, we show that visualisations of SHAP results can be used to identify attack-specific artefacts and the differences and consistencies between synthetic speech and converted voice spoofing attacks.

eess.AS↗

arXiv2021

Explaining deep learning models for spoofing and deepfake detection with SHapley Additive exPlanations

Substantial progress in spoofing and deepfake detection has been made in recent years. Nonetheless, the community has yet to make notable inroads in providing an explanation for how a classifier produces its output. The dominance of black box spoofing detection solutions is at further odds with the drive toward trustworthy, explainable artificial intelligence. This paper describes our use of SHapley Additive exPlanations (SHAP) to gain new insights in spoofing detection. We demonstrate use of the tool in revealing unexpected classifier behaviour, the artefacts that contribute most to classifier outputs and differences in the behaviour of competing spoofing detection models. The tool is both efficient and flexible, being readily applicable to a host of different architecture models in addition to related, different applications. All results reported in the paper are reproducible using open-source software.

eess.AS↗

arXiv2021

Raw Differentiable Architecture Search for Speech Deepfake and Spoofing Detection

End-to-end approaches to anti-spoofing, especially those which operate directly upon the raw signal, are starting to be competitive with their more traditional counterparts. Until recently, all such approaches consider only the learning of network parameters; the network architecture is still hand crafted. This too, however, can also be learned. Described in this paper is our attempt to learn automatically the network architecture of a speech deepfake and spoofing detection solution, while jointly optimising other network components and parameters, such as the first convolutional layer which operates on raw signal inputs. The resulting raw differentiable architecture search system delivers a tandem detection cost function score of 0.0517 for the ASVspoof 2019 logical access database, a result which is among the best single-system results reported to date.

eess.AS↗