Beyond Neural-on-Neural Approaches to Speaker Gender Protection

TL;DR

This paper advocates for a broader evaluation of speaker gender protection methods by incorporating speech science features, human vocal adversaries, and traditional speech analysis to improve privacy and interpretability.

Contribution

It introduces the importance of using speech features and human vocal adversaries alongside neural methods for more comprehensive gender protection evaluation.

Findings

Neural classifiers alone may not fully assess gender protection effectiveness.

Speech features from speech scientists provide valuable insights into protection mechanisms.

Human vocal adversaries can serve as interpretable benchmarks for gender protection.

Abstract

Recent research has proposed approaches that modify speech to defend against gender inference attacks. The goal of these protection algorithms is to control the availability of information about a speaker's gender, a privacy-sensitive attribute. Currently, the common practice for developing and testing gender protection algorithms is "neural-on-neural", i.e., perturbations are generated and tested with a neural network. In this paper, we propose to go beyond this practice to strengthen the study of gender protection. First, we demonstrate the importance of testing gender inference attacks that are based on speech features historically developed by speech scientists, alongside the conventionally used neural classifiers. Next, we argue that researchers should use speech features to gain insight into how protective modifications change the speech signal. Finally, we point out that gender-protection algorithms should be compared with novel "vocal adversaries", human-executed voice adaptations, in order to improve interpretability and enable before-the-mic protection.