DAST: A Dual-Stream Voice Anonymization Attacker with Staged Training

• URL: http://arxiv.org/abs/2603.12840v2
• Date: Mon, 16 Mar 2026 03:47:49 GMT
• Title: DAST: A Dual-Stream Voice Anonymization Attacker with Staged Training
• Abstract summary: We propose a dual-stream attacker that fuses spectral and self-supervised learning features via parallel encoders.<n>We show that Stage II is the primary driver of generalization, enabling strong attacking performance on unseen anonymization datasets.<n>With Stage III, fine-tuning on only 10% of the target anonymization dataset surpasses current state-of-the-art attackers in terms of EER.
• Score: 20.576907343629188
• License: http://creativecommons.org/licenses/by-nc-nd/4.0/
• Abstract: Voice anonymization masks vocal traits while preserving linguistic content, which may still leak speaker-specific patterns. To assess and strengthen privacy evaluation, we propose a dual-stream attacker that fuses spectral and self-supervised learning features via parallel encoders with a three-stage training strategy. Stage I establishes foundational speaker-discriminative representations. Stage II leverages the shared identity-transformation characteristics of voice conversion and anonymization, exposing the model to diverse converted speech to build cross-system robustness. Stage III provides lightweight adaptation to target anonymized data. Results on the VoicePrivacy Attacker Challenge (VPAC) dataset demonstrate that Stage II is the primary driver of generalization, enabling strong attacking performance on unseen anonymization datasets. With Stage III, fine-tuning on only 10\% of the target anonymization dataset surpasses current state-of-the-art attackers in terms of EER.

Related papers

• Hearing the Whispers: Black-Box Membership Inference Attacks on Finetuned TTS Models [23.501699222486412]
  Existing black-box membership inference attacks (MIAs) face unique challenges when adapted to Text-to-Speech (TTS) models.<n>We present the first black-box MIA framework explicitly tailored to TTS models at both the speaker and record levels.
  arXiv  Detail & Related papers  (2026-09-01T18:00:45Z)
• Taming Text-to-Sounding Video Generation via Advanced Modality Condition and Interaction [28.20791917022439]
  This study focuses on a challenging yet promising task, Text-to-Sounding-Video (T2SV) generation.<n>It aims to generate a video with synchronized audio from text conditions, ensuring both modalities are aligned with text.<n>Two critical challenges still remain unaddressed: (1) a single, shared text caption where the text for video is equal to the text for audio often creates modal interference, and (2) the optimal mechanism for cross-modal feature interaction remains unclear.
  arXiv  Detail & Related papers  (2025-10-03T15:43:56Z)
• Backdoor Attacks Against Speech Language Models [63.07317091368079]
  We present the first systematic study of audio backdoor attacks against speech language models.<n>We demonstrate its effectiveness across four speech encoders and three datasets, covering four tasks.<n>We propose a fine-tuning-based defense that mitigates the threat of poisoned pretrained encoders.
  arXiv  Detail & Related papers  (2025-10-01T17:45:04Z)
• Multi-speaker Text-to-speech Training with Speaker Anonymized Data [40.70515431989197]
  We investigate training multi-speaker text-to-speech (TTS) models using data that underwent speaker anonymization (SA) Two signal processing-based and three deep neural network-based SA methods were used to anonymize VCTK, a multi-speaker TTS dataset. We conducted extensive objective and subjective experiments to evaluate the anonymized training data, as well as the performance of the downstream TTS model trained using those data.
  arXiv  Detail & Related papers  (2024-05-20T03:55:44Z)
• Voice Anonymization for All -- Bias Evaluation of the Voice Privacy Challenge Baseline System [0.48342038441006807]
  This study investigates bias in voice anonymization systems within the context of the Voice Privacy Challenge. We curate a novel benchmark dataset to assess performance disparities among speaker subgroups based on sex and dialect.
  arXiv  Detail & Related papers  (2023-11-27T13:26:49Z)
• Speech-to-Speech Translation with Discrete-Unit-Based Style Transfer [53.72998363956454]
  Direct speech-to-speech translation (S2ST) with discrete self-supervised representations has achieved remarkable accuracy. The scarcity of high-quality speaker-parallel data poses a challenge for learning style transfer during translation. We design an S2ST pipeline with style-transfer capability on the basis of discrete self-supervised speech representations and timbre units.
  arXiv  Detail & Related papers  (2023-09-14T09:52:08Z)
• AV-data2vec: Self-supervised Learning of Audio-Visual Speech Representations with Contextualized Target Representations [88.30635799280923]
  We introduce AV-data2vec which builds audio-visual representations based on predicting contextualized representations. Results on LRS3 show that AV-data2vec consistently outperforms existing methods with the same amount of data and model size.
  arXiv  Detail & Related papers  (2023-02-10T02:55:52Z)
• V-Cloak: Intelligibility-, Naturalness- & Timbre-Preserving Real-Time Voice Anonymization [0.0]
  We develop a voice anonymization system, named V-Cloak, which attains real-time voice anonymization. Our designed anonymizer features a one-shot generative model that modulates the features of the original audio at different frequency levels. Experiment results confirm that V-Cloak outperforms five baselines in terms of anonymity performance.
  arXiv  Detail & Related papers  (2022-10-27T02:58:57Z)
• The VoicePrivacy 2022 Challenge Evaluation Plan [46.807999940446294]
  Training, development and evaluation datasets are provided. Participants apply their developed anonymization systems. Results will be presented at a workshop held in conjunction with INTERSPEECH 2022.
  arXiv  Detail & Related papers  (2022-03-23T15:05:18Z)
• Direct speech-to-speech translation with discrete units [64.19830539866072]
  We present a direct speech-to-speech translation (S2ST) model that translates speech from one language to speech in another language without relying on intermediate text generation. We propose to predict the self-supervised discrete representations learned from an unlabeled speech corpus instead. When target text transcripts are available, we design a multitask learning framework with joint speech and text training that enables the model to generate dual mode output (speech and text) simultaneously in the same inference pass.
  arXiv  Detail & Related papers  (2021-07-12T17:40:43Z)
• Disentangled Speech Embeddings using Cross-modal Self-supervision [119.94362407747437]
  We develop a self-supervised learning objective that exploits the natural cross-modal synchrony between faces and audio in video. We construct a two-stream architecture which: (1) shares low-level features common to both representations; and (2) provides a natural mechanism for explicitly disentangling these factors.
  arXiv  Detail & Related papers  (2020-02-20T14:13:12Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.