Non-Autoregressive TTS with Explicit Duration Modelling for Low-Resource Highly Expressive Speech

TL;DR

This paper introduces a non-autoregressive, explicit duration modeling approach for low-resource, highly expressive speech synthesis, significantly improving naturalness and speaker similarity with only 15 minutes of target data.

Contribution

It presents a novel non-autoregressive TTS method with explicit duration modeling and cGAN fine-tuning, enabling high-quality expressive speech synthesis from minimal data.

Findings

Achieves 23.3% improvement in naturalness over previous methods

Matches full-data Tacotron2 performance with only 15 minutes of data

Outperforms full-data models with 30+ minutes of target speech

Abstract

Whilst recent neural text-to-speech (TTS) approaches produce high-quality speech, they typically require a large amount of recordings from the target speaker. In previous work, a 3-step method was proposed to generate high-quality TTS while greatly reducing the amount of data required for training. However, we have observed a ceiling effect in the level of naturalness achievable for highly expressive voices when using this approach. In this paper, we present a method for building highly expressive TTS voices with as little as 15 minutes of speech data from the target speaker. Compared to the current state-of-the-art approach, our proposed improvements close the gap to recordings by 23.3% for naturalness of speech and by 16.3% for speaker similarity. Further, we match the naturalness and speaker similarity of a Tacotron2-based full-data (~10 hours) model using only 15 minutes of target speaker data, whereas with 30 minutes or more, we significantly outperform it. The following improvements are proposed: 1) changing from an autoregressive, attention-based TTS model to a non-autoregressive model replacing attention with an external duration model and 2) an additional Conditional Generative Adversarial Network (cGAN) based fine-tuning step.