Do Not Mimic My Voice: Teacher-Guided Unlearning for Zero-Shot Text-to-Speech

Taesoo Kim, Jinju Kim, Dong Chan Kim, Jong Hwan Ko, Gyeong-Moon Park

Keywords: zero-shot tts, machine unlearning, voice privacy

TL;DR: Enhancing voice privacy through machine unlearning in zero-shot text-to-speech

Abstract: The rapid advancement of Zero-Shot Text-to-Speech (ZS-TTS) technology has enabled high-fidelity voice synthesis from minimal audio cues, raising significant privacy and ethical concerns. In particular, the ability to replicate an individual’s voice without consent poses risks, highlighting the need for machine unlearning techniques to protect voice privacy. In this paper, we introduce the first machine unlearning framework for ZS-TTS, Teacher-Guided Unlearning (TGU), designed to ensure that the model forgets designated speaker identities while retaining its ability to generate accurate speech for other speakers. Unlike conventional unlearning methods, TGU leverages randomness to prevent consistent replication of forget speakers' voices, ensuring unlearned identities remain untraceable. Additionally, we propose a new evaluation metric, speaker-Zero Retrain Forgetting (spk-ZRF), which measures the model’s effectiveness in preventing the reproduction of forgotten voices. The experiments conducted on the state-of-the-art model demonstrate that TGU prevents the model from replicating forget speakers' voices while maintaining high quality for other speakers. The demo is available at https://speechunlearn.github.io/

Primary Area: alignment, fairness, safety, privacy, and societal considerations

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.

Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Submission Number: 9378