Beyond EER: Multi-Dimensional Evaluation of Information Leakage in Speaker De-Identification

2026-09-16

AI Summary

本文针对说话人去识别化中信息泄露的多维度评估问题，提出了一种包含五个互补指标的综合评估框架，以更全面地衡量隐私保护效果。

Abstract

Speaker de-identification (SDID) aims to preserve privacy by concealing speaker identity while maintaining speech utility. However, current evaluations often reduce privacy to a single dimension - biometric verification performance - typically measured by Equal Error Rate (EER). This narrow focus ignores critical leakage channels, such as soft biometric inference, embedding-level re-identification, and structural template similarity, which threaten the unlinkability and irreversibility of biometric references. We propose a holistic evaluation framework across five complementary metrics: (i) EER, (ii) soft biometric leakage score , (iii) cumulative match characteristic re-identification analysis, (iv) canonical correlation analysis and Procrustes embedding alignment, and (v) intelligibility via word error rate and semantic similarity. Evaluating five SDID systems from the IARPA ARTS program, we demonstrate that these metrics capture independent dimensions of information leakage. Our results indicate that reliance on a single metric can misrepresent the privacy properties of an SDID system.

Problem

• Speaker De-Identification
• Information Leakage
• Biometric Verification
• Soft Biometrics
• Re-identification

Innovation

• multi-dimensional evaluation
• information leakage
• speaker de-identification
• privacy protection
• biometric verification