Simple and Effective Multi-sentence TTS with Expressive and Coherent Prosody

TL;DR

This paper enhances multi-sentence TTS by extending Transformer-based models with long context, rich text features, and multi-speaker training, leading to more expressive, coherent, and natural speech.

Contribution

It introduces simple yet effective extensions to a FastSpeech-like TTS system, leveraging long context and multi-speaker data to improve prosody and naturalness.

Findings

Long context improves prosody coherence.

Multi-speaker training enhances expressiveness.

Proposed system outperforms competitors in naturalness.

Abstract

Generating expressive and contextually appropriate prosody remains a challenge for modern text-to-speech (TTS) systems. This is particularly evident for long, multi-sentence inputs. In this paper, we examine simple extensions to a Transformer-based FastSpeech-like system, with the goal of improving prosody for multi-sentence TTS. We find that long context, powerful text features, and training on multi-speaker data all improve prosody. More interestingly, they result in synergies. Long context disambiguates prosody, improves coherence, and plays to the strengths of Transformers. Fine-tuning word-level features from a powerful language model, such as BERT, appears to profit from more training data, readily available in a multi-speaker setting. We look into objective metrics on pausing and pacing and perform thorough subjective evaluations for speech naturalness. Our main system, which incorporates all the extensions, achieves consistently strong results, including statistically significant improvements in speech naturalness over all its competitors.