PPSpeech: Phrase based Parallel End-to-End TTS System

TL;DR

PPSpeech is a novel TTS system that combines phrase-based parallel processing with autoregressive modeling within phrases, achieving high-quality, stable, and efficient speech synthesis suitable for online applications.

Contribution

The paper introduces PPSpeech, a phrase-based parallel end-to-end TTS system that improves synthesis speed and style stability over traditional autoregressive models.

Findings

PPSpeech significantly outperforms Tacotron 2 in synthesis speed for longer sentences.

The system maintains stable style and timbre across sentences.

Subjective evaluations favor PPSpeech over GST in naturalness and style transition.

Abstract

Current end-to-end autoregressive TTS systems (e.g. Tacotron 2) have outperformed traditional parallel approaches on the quality of synthesized speech. However, they introduce new problems at the same time. Due to the autoregressive nature, the time cost of inference has to be proportional to the length of text, which pose a great challenge for online serving. On the other hand, the style of synthetic speech becomes unstable and may change obviously among sentences. In this paper, we propose a Phrase based Parallel End-to-End TTS System (PPSpeech) to address these issues. PPSpeech uses autoregression approach within a phrase and executes parallel strategies for different phrases. By this method, we can achieve both high quality and high efficiency. In addition, we propose acoustic embedding and text context embedding as the conditions of encoder to keep successive and prevent from abrupt style or timbre change. Experiments show that, the synthesis speed of PPSpeech is much faster than sentence level autoregressive Tacotron 2 when a sentence has more than 5 phrases. The speed advantage increases with the growth of sentence length. Subjective experiments show that the proposed system with acoustic embedding and context embedding as conditions can make the style transition across sentences gradient and natural, defeating Global Style Token (GST) obviously in MOS.