English

Single-stage TTS with Masked Audio Token Modeling and Semantic Knowledge Distillation

Sound 2024-09-18 v1 Artificial Intelligence Audio and Speech Processing Signal Processing

Abstract

Audio token modeling has become a powerful framework for speech synthesis, with two-stage approaches employing semantic tokens remaining prevalent. In this paper, we aim to simplify this process by introducing a semantic knowledge distillation method that enables high-quality speech generation in a single stage. Our proposed model improves speech quality, intelligibility, and speaker similarity compared to a single-stage baseline. Although two-stage systems still lead in intelligibility, our model significantly narrows the gap while delivering comparable speech quality. These findings showcase the potential of single-stage models to achieve efficient, high-quality TTS with a more compact and streamlined architecture.

Keywords

Cite

@article{arxiv.2409.11003,
  title  = {Single-stage TTS with Masked Audio Token Modeling and Semantic Knowledge Distillation},
  author = {Gerard I. Gállego and Roy Fejgin and Chunghsin Yeh and Xiaoyu Liu and Gautam Bhattacharya},
  journal= {arXiv preprint arXiv:2409.11003},
  year   = {2024}
}

Comments

Demo page: see https://narsistts.github.io

R2 v1 2026-06-28T18:47:33.256Z