Nonparallel Emotional Voice Conversion For Unseen Speaker-Emotion Pairs Using Dual Domain Adversarial Network & Virtual Domain Pairing

TL;DR

This paper introduces a novel emotional voice conversion method that enables conversion for unseen speaker-emotion pairs by extending a GAN-based architecture with dual encoders and a virtual domain pairing strategy, demonstrated on Hindi data.

Contribution

It proposes a dual encoder architecture with virtual domain pairing to convert emotions for unseen speaker-emotion combinations in voice conversion.

Findings

Effective conversion for unseen speaker-emotion pairs demonstrated

Uses dual encoders and virtual domain pairing to improve generalization

Evaluated successfully on Hindi emotional speech database

Abstract

Primary goal of an emotional voice conversion (EVC) system is to convert the emotion of a given speech signal from one style to another style without modifying the linguistic content of the signal. Most of the state-of-the-art approaches convert emotions for seen speaker-emotion combinations only. In this paper, we tackle the problem of converting the emotion of speakers whose only neutral data are present during the time of training and testing (i.e., unseen speaker-emotion combinations). To this end, we extend a recently proposed StartGANv2-VC architecture by utilizing dual encoders for learning the speaker and emotion style embeddings separately along with dual domain source classifiers. For achieving the conversion to unseen speaker-emotion combinations, we propose a Virtual Domain Pairing (VDP) training strategy, which virtually incorporates the speaker-emotion pairs that are not present in the real data without compromising the min-max game of a discriminator and generator in adversarial training. We evaluate the proposed method using a Hindi emotional database.