Kokoro TTS

A local implementation of the Kokoro Text-to-Speech system, based on the Kokoro-82M model.

Features

• High-quality English text-to-speech synthesis
• Multiple voice styles
• Adjustable speech speed
• Local inference without internet dependency (after initial model download)
• Automatic espeak-ng installation using espeakng-loader

Prerequisites

• Python 3.8 or higher
• Git LFS (for model download)

Installation

1. Set up Python environment:

# Create and activate virtual environment
python -m venv venv
.\venv\Scripts\activate  # Windows
source venv/bin/activate  # Linux/Mac

# Install dependencies
pip install -r requirements.txt

The espeakng-loader package will automatically handle the installation of espeak-ng.

Project Structure

• tts_demo.py - Demo script showing basic usage
• models.py - Model implementation and utilities
• requirements.txt - Python dependencies

Usage

Run the demo script:

python tts_demo.py

This will:

1. Automatically install espeak-ng if needed
2. Download the Kokoro model
3. Generate a sample audio file

Credits

• Original model: hexgrad/Kokoro-82M
• espeak-ng loader: espeakng-loader
• Based on papers:
  • arxiv: 2306.07691
  • arxiv: 2203.02395

License

Apache-2.0 License (following the original model's license)