StyleTTS 2 - lite

A lightweight, efficient variation of the StyleTTS 2 text‐to‐speech model, optimized for rapid integration into your applications. With a compact 90 million parameter footprint and built‐in speaker‑and‑language tagging, you can seamlessly switch voices and languages even within a single sentence.

Online Demo

Explore the model on Hugging Face Spaces:
https://huggingface.co/spaces/dangtr0408/StyleTTS2-lite-vi-space

Fine-tune

https://github.com/dangtr0408/StyleTTS2-lite

Training Details

Base Checkpoint:
- Initialized from the official StyleTTS 2 LibriTTS weights.
Token Extension:
- Expanded the token set to 189 symbols to ensure full Vietnamese IPA compatibility.
Training Data:
- FonosVietnam (extracted from the viVoice corpus)
- VoizFM (extracted from the viVoice corpus)
Training Schedule:
- Trained for 120 000 steps.

Model Architecture

Component	Parameters
Decoder	54 ,289 ,492
Predictor	16 ,194 ,612
Style Encoder	13 ,845 ,440
Text Encoder	5,612 ,320
Total	89 ,941 ,576

Prerequisites

Python: Version 3.7 or higher
Git: To clone the repository

Installation & Setup

Clone the repository


git  clone  https://huggingface.co/dangtr0408/StyleTTS2-lite-vi

cd  StyleTTS2-lite-vi

Install dependencies:


pip  install  -r  requirements.txt

On Linux, manually install espeak:


sudo  apt-get  install  espeak-ng

Usage Example

See run.ipynb file.

Disclaimer

Before using these pre-trained models, you agree to inform the listeners that the speech samples are synthesized by the pre-trained models, unless you have the permission to use the voice you synthesize. That is, you agree to only use voices whose speakers grant the permission to have their voice cloned, either directly or by license before making synthesized voices public, or you have to publicly announce that these voices are synthesized if you do not have the permission to use these voices.

References

License

Code: MIT License

Model: CC-BY-NC-SA-4.0

dangtr0408
/

StyleTTS2-lite-vi