mlsanityData to 14 Oct 2026

Evaluating Identity Leakage in Speaker De-Identification Systems

arXiv 2508.14012 ↗First posted 19 Aug 2025cs.SD cs.AIcited by 0

Not on the map. Why?

Abstract (arXiv)

Speaker de-identification aims to conceal a speaker's identity while preserving intelligibility of the underlying speech. We introduce a benchmark that quantifies residual identity leakage with three complementary error rates: equal error rate, cumulative match characteristic hit rate, and embedding-space similarity measured via canonical correlation analysis and Procrustes analysis. Evaluation results reveal that all state-of-the-art speaker de-identification systems leak identity information. The highest performing system in our evaluation performs only slightly better than random guessing, while the lowest performing system achieves a 45% hit rate within the top 50 candidates based on CMC. These findings highlight persistent privacy risks in current speaker de-identification technologies.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 12 of its 22 references that are in the map. First posted on arXiv 19 Aug 2025.

Builds on

12 of 22 references.

References · 12

  1. 2021SoundStream: An End-to-End Neural Audio Codeccited by 771
  2. 2019AUTOVC: Zero-Shot Voice Style Transfer with Only Autoencoder Losscited by 192
  3. 2021TitaNet: Neural Model for speaker representation with 1D Depth-wise separable convolutions and global contextcited by 62
  4. 2021ECAPA-TDNN Embeddings for Speaker Diarizationcited by 47
  5. 2024HLTCOE JHU Submission to the Voice Privacy Challenge 2024cited by 10
  6. 2024Analysis of Speech Temporal Dynamics in the Context of Speaker Verification and Voice Anonymizationcited by 5
  7. 2024Attacking Voice Anonymization Systems with Augmented Feature and Speaker Identity Differencecited by 5
  8. 2025SpecWav-Attack: Leveraging Spectrogram Resizing and Wav2Vec 2.0 for Attacking Anonymized Speechcited by 5
  9. 2025Exploiting Context-dependent Duration Features for Voice Anonymization Attack Systemscited by 3
  10. 2025DiffAttack: Diffusion-based Timbre-reserved Adversarial Attack in Speaker Identificationcited by 2
  11. 2024Toward Improving Synthetic Audio Spoofing Detection Robustness via Meta-Learning and Disentangled Training With Adversarial Examplescited by 1
  12. 2025UniPET-SPK: A Unified Framework for Parameter-Efficient Tuning of Pre-trained Speech Models for Robust Speaker Verificationcited by 1

Built on by

Not cited yet.