A Non-intrusive and Adaptive Speaker De-Identification Scheme Using Adversarial Examples
Published in The 28th Annual Conference on Mobile Computing and Networking (ACM MobiCom 2022), 2022
Recommended citation: Meng Chen, Li Lu*, Jiadi Yu, Yingying Chen, Zhongjie Ba, Feng Lin, Kui Ren. "A Non-intrusive and Adaptive Speaker De-Identification Scheme Using Adversarial Examples." Proceedings of the 28th Annual Conference on Mobile Computing and Networking (ACM MobiCom 2022). Sydney, Australia. pp. 853-855. 2022. doi: 10.1145/3495243.3558260.
This work received the Best Poster Runner-up Award of ACM MobiCom 2022.
This work was presented as a poster in ACM MobiCom 2022. ACM Annual Conference on Mobile Computing and Networking is a premier international conference in mobile computing and communication areas. ACM MobiCom is also a CCF-A conference.
Abstract: Faced with the threat of identity leakage during voice data publishing, users are engaged in a privacy-utility dilemma while enjoying convenient voice services. Existing studies employ direct modification or text-based re-synthesis to de-identify users’ voices, but resulting in inconsistent audibility for human participants and not adaptive to informed attacks. In this poster, we propose a non-intrusive and adaptive speaker de-identification scheme to balance the privacy and utility of voice services. We generate adversarial examples to conceal user identity from exposure by Automatic Speaker Identification (ASI). By learning a compact distribution with a conditional variational auto-encoder, our system enables on-demand target sampling and diverse identity transformation. We also introduce the acoustic masking effect to construct inaudible perturbations, thus preserving the speech content and perceptual quality.