Voice Conversion-based Privacy through Adversarial Information Hiding

TL;DR

This paper introduces a privacy-preserving voice conversion method using adversarial information hiding, enabling control over identity leakage while maintaining speech content, improving privacy and flexibility over existing techniques.

Contribution

The paper proposes a novel adversarial information hiding approach for voice conversion that balances speaker privacy with speech content preservation, surpassing prior methods in privacy control.

Findings

Successfully modifies perceived speaker identity

Maintains source lexical content effectively

Offers flexible privacy-utility trade-offs

Abstract

Privacy-preserving voice conversion aims to remove only the attributes of speech audio that convey identity information, keeping other speech characteristics intact. This paper presents a mechanism for privacy-preserving voice conversion that allows controlling the leakage of identity-bearing information using adversarial information hiding. This enables a deliberate trade-off between maintaining source-speech characteristics and modification of speaker identity. As such, the approach improves on voice-conversion techniques like CycleGAN and StarGAN, which were not designed for privacy, meaning that converted speech may leak personal information in unpredictable ways. Our approach is also more flexible than ASR-TTS voice conversion pipelines, which by design discard all prosodic information linked to textual content. Evaluations show that the proposed system successfully modifies perceived speaker identity whilst well maintaining source lexical content.