Training Text-To-Speech Systems From Synthetic Data: A Practical Approach For Accent Transfer Tasks

TL;DR

This paper presents a practical method for accent transfer in text-to-speech systems by training robust models on synthetic data generated from less stable transfer models, achieving high-quality, accent-transferred speech.

Contribution

It introduces a novel approach of using synthetic data from a less robust TTS system to train a more stable and high-quality accent transfer TTS model.

Findings

Models trained on synthetic data can produce high-quality accent transfer speech.

The approach preserves speaker characteristics like speaking style.

Synthetic data training enables practical and stable TTS accent transfer.

Abstract

Transfer tasks in text-to-speech (TTS) synthesis - where one or more aspects of the speech of one set of speakers is transferred to another set of speakers that do not feature these aspects originally - remains a challenging task. One of the challenges is that models that have high-quality transfer capabilities can have issues in stability, making them impractical for user-facing critical tasks. This paper demonstrates that transfer can be obtained by training a robust TTS system on data generated by a less robust TTS system designed for a high-quality transfer task; in particular, a CHiVE-BERT monolingual TTS system is trained on the output of a Tacotron model designed for accent transfer. While some quality loss is inevitable with this approach, experimental results show that the models trained on synthetic data this way can produce high quality audio displaying accent transfer, while preserving speaker characteristics such as speaking style.