Evaluating X-vector-based Speaker Anonymization under White-box Assessment
This work addresses privacy concerns in voice data by providing a more rigorous assessment of anonymization techniques, though it appears incremental as it builds on existing Voice Privacy challenge frameworks.
The paper tackled the problem of evaluating speaker anonymization by constraining target identity selection to a specific identity instead of random, under a white-box attack scenario, and found that this approach allows investigation of which target identities are more effective for anonymizing specific speakers.
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.