Evaluating X-vector-based Speaker Anonymization under White-box Assessment

TL;DR

This paper assesses the robustness of X-vector-based speaker anonymization when the attacker has complete knowledge of the system, focusing on fixed target identities and their effectiveness in protecting speaker privacy.

Contribution

It introduces a white-box evaluation framework by fixing target identities, enabling analysis of the impact of specific targets on anonymization effectiveness.

Findings

Certain target identities provide better anonymization.

Fixed target selection reveals vulnerabilities in the anonymization process.

White-box assessment exposes potential privacy risks.

Abstract

In the scenario of the Voice Privacy challenge, anonymization is achieved by converting all utterances from a source speaker to match the same target identity; this identity being randomly selected. In this context, an attacker with maximum knowledge about the anonymization system can not infer the target identity. This article proposed to constrain the target selection to a specific identity, i.e., removing the random selection of identity, to evaluate the extreme threat under a whitebox assessment (the attacker has complete knowledge about the system). Targeting a unique identity also allows us to investigate whether some target's identities are better than others to anonymize a given speaker.