Yuna TTS

Welcome to Yuna TTS, a leading text-to-speech (TTS) engine designed to provide a natural and expressive voice for your applications. Yuna TTS is a deep learning-based TTS engine that can generate high-quality speech from text. It is designed to be easy to use and integrate into your applications, with support for multiple languages and voices.

Table of Content

• Yuna TTS
  • Table of Content
  • Installation
  • Usage
    • Language Support
    • CLI
    • Python API
  • Training
    • Data Preparation
    • Training
    • Inference

Installation

The repo is developed and tested on Ubuntu 20.04 and Python 3.9. To install YunaTTS, you can follow the instructions below:

git clone https://github.com/yukiarimo/yunatts.git
cd yunatts
pip install -e .
python -m unidic download

Usage

Language Support

Languages supported by Yuna TTS:

• English (EN)
• Spanish (ES)
• French (FR)
• Chinese (ZH)
• Japanese (JP)
• Korean (KR)

CLI

You may use the YunaTTS CLI to interact with YunaTTS. The CLI may be invoked using either yunatts or yuna. Here are some examples:

Read English text:

yuna "Text to read" output.wav

Specify a language:

yuna "Text to read" output.wav --language EN

Specify a speaker:

yuna "Text to read" output.wav --language EN --speaker EN-US
yuna "Text to read" output.wav --language EN --speaker EN-AU

Specify a speed:

yuna "Text to read" output.wav --language EN --speaker EN-US --speed 1.5
yuna "Text to read" output.wav --speed 1.5

Load from a file:

yuna file.txt out.wav --file

Python API

from yuna.api import TTS

# Speed is adjustable
speed = 1.0

# CPU is sufficient for real-time inference.
# You can set it manually to 'cpu' or 'cuda' or 'cuda:0' or 'mps'
device = 'auto' # Will automatically use GPU if available
language = 'EN'

# English 
text = "Did you ever hear a folk tale about a giant turtle?"
model = TTS(language=language, device=device)
output_path = 'en.wav'
model.tts_to_file(text, 0, output_path, speed=speed)

Training

Before training, please install YunaTTS in dev mode and go to the yuna folder.

pip install -e .
cd yuna

Data Preparation

To train a TTS model, we need to prepare the audio files and a metadata file.

path/to/audio_001.wav|<speaker_name>|<language_code>|<text_001>
path/to/audio_002.wav|<speaker_name>|<language_code>|<text_002>

We can then run the preprocessing code:

python preprocess_text.py --metadata data/example/metadata.list 

A config file data/example/config.json will be generated. Feel free to edit some hyper-parameters in that config file (for example, you may decrease the batch size if you have encountered the CUDA out-of-memory issue).

Training

The training can be launched by:

bash train.sh <path/to/config.json> <num_of_gpus>

Inference

Simply run:

python infer.py --text "<some text here>" -m /path/to/checkpoint/G_<iter>.pth -o <output_dir>