Exploring the Importance of F0 Trajectories for Speaker Anonymization using X-vectors and Neural Waveform Models

TL;DR

This paper investigates the role of F0 trajectories in speaker anonymization, demonstrating that F0 modifications can significantly enhance anonymization effectiveness with minimal impact on speech recognition accuracy.

Contribution

It introduces and evaluates eight low-complexity F0 modification methods within a speaker anonymization framework, highlighting the importance of F0 in privacy preservation.

Findings

F0 modifications can improve anonymization by up to 8%.

F0 adjustments cause minor word-error rate degradation.

F0 plays a crucial role in speaker anonymization effectiveness.

Abstract

Voice conversion for speaker anonymization is an emerging field in speech processing research. Many state-of-the-art approaches are based on the resynthesis of the phoneme posteriorgrams (PPG), the fundamental frequency (F0) of the input signal together with modified X-vectors. Our research focuses on the role of F0 for speaker anonymization, which is an understudied area. Utilizing the VoicePrivacy Challenge 2020 framework and its datasets we developed and evaluated eight low-complexity F0 modifications prior resynthesis. We found that modifying the F0 can improve speaker anonymization by as much as 8% with minor word-error rate degradation.