Hierarchical and Multi-Scale Variational Autoencoder for Diverse and Natural Non-Autoregressive Text-to-Speech

TL;DR

This paper introduces HiMuV-TTS, a hierarchical variational autoencoder-based non-autoregressive TTS model that enhances speech naturalness and diversity by modeling prosody at multiple scales with adversarial training.

Contribution

The paper presents a novel hierarchical multi-scale VAE framework for NAR-TTS that improves speech diversity and naturalness over single-scale models.

Findings

Generated speech is more diverse and natural.

Model effectively captures prosody at different scales.

Outperforms existing single-scale VAE TTS models.

Abstract

This paper proposes a hierarchical and multi-scale variational autoencoder-based non-autoregressive text-to-speech model (HiMuV-TTS) to generate natural speech with diverse speaking styles. Recent advances in non-autoregressive TTS (NAR-TTS) models have significantly improved the inference speed and robustness of synthesized speech. However, the diversity of speaking styles and naturalness are needed to be improved. To solve this problem, we propose the HiMuV-TTS model that first determines the global-scale prosody and then determines the local-scale prosody via conditioning on the global-scale prosody and the learned text representation. In addition, we improve the quality of speech by adopting the adversarial training technique. Experimental results verify that the proposed HiMuV-TTS model can generate more diverse and natural speech as compared to TTS models with single-scale variational autoencoders, and can represent different prosody information in each scale.