Generating Multilingual Gender-Ambiguous Text-to-Speech Voices

TL;DR

This paper introduces a novel method for generating diverse, gender-ambiguous multilingual TTS voices by sampling from a latent speaker space, with evaluations showing increased ambiguity and robustness across listener demographics.

Contribution

It presents the first systematic approach to reliably generate a variety of gender-ambiguous voices in multiple languages using a gender-aware sampling method.

Findings

Generated voices are perceived as more gender-ambiguous than baseline.

The method produces diverse and consistent voices across languages.

Gender perception remains robust across listener demographics.

Abstract

The gender of any voice user interface is a key element of its perceived identity. Recently, there has been increasing interest in interfaces where the gender is ambiguous rather than clearly identifying as female or male. This work addresses the task of generating novel gender-ambiguous TTS voices in a multi-speaker, multilingual setting. This is accomplished by efficiently sampling from a latent speaker embedding space using a proposed gender-aware method. Extensive objective and subjective evaluations clearly indicate that this method is able to efficiently generate a range of novel, diverse voices that are consistent and perceived as more gender-ambiguous than a baseline voice across all the languages examined. Interestingly, the gender perception is found to be robust across two demographic factors of the listeners: native language and gender. To our knowledge, this is the first systematic and validated approach that can reliably generate a variety of gender-ambiguous voices.