Cross-Speaker Emotion Transfer for Low-Resource Text-to-Speech Using Non-Parallel Voice Conversion with Pitch-Shift Data Augmentation

TL;DR

This paper introduces a novel data augmentation approach combining pitch-shifting and voice conversion to improve low-resource expressive TTS, achieving better naturalness and emotional similarity with limited data.

Contribution

It proposes a new pitch-shift and VC-based data augmentation method that stabilizes training and enhances emotional TTS quality in low-resource scenarios.

Findings

Improved naturalness in emotional TTS.

Enhanced emotional similarity according to subjective tests.

Effective stabilization of VC and TTS training with limited data.

Abstract

Data augmentation via voice conversion (VC) has been successfully applied to low-resource expressive text-to-speech (TTS) when only neutral data for the target speaker are available. Although the quality of VC is crucial for this approach, it is challenging to learn a stable VC model because the amount of data is limited in low-resource scenarios, and highly expressive speech has large acoustic variety. To address this issue, we propose a novel data augmentation method that combines pitch-shifting and VC techniques. Because pitch-shift data augmentation enables the coverage of a variety of pitch dynamics, it greatly stabilizes training for both VC and TTS models, even when only 1,000 utterances of the target speaker's neutral data are available. Subjective test results showed that a FastSpeech 2-based emotional TTS system with the proposed method improved naturalness and emotional similarity compared with conventional methods.