Speaker Identification Experiments Under Gender De-Identification

TL;DR

This study evaluates how different voice modification algorithms affect gender recognition accuracy, focusing on the extent, quality, and reversibility of speech tone changes to improve privacy in multimedia content.

Contribution

It introduces an analysis of four voice modification algorithms' impact on gender recognition, assessing their effectiveness and reversibility for de-identification purposes.

Findings

Certain algorithms significantly reduce gender recognition accuracy.

Modifications vary in reversibility and speech quality.

Some methods balance privacy and speech intelligibility effectively.

Abstract

The present work is based on the COST Action IC1206 for De-identification in multimedia content. It was performed to test four algorithms of voice modifications on a speech gender recognizer to find the degree of modification of pitch when the speech recognizer have the probability of success equal to the probability of failure. The purpose of this analysis is to assess the intensity of the speech tone modification, the quality, the reversibility and not-reversibility of the changes made. Keywords DeIdentification; Speech Algorithms