Neo TTS

A modern, modular Text-to-Speech (TTS) system built with Flask, featuring local inference with Kokoro TTS models.

🚀 Features

• Modular Architecture: Easily extensible with new TTS models
• Web Interface: Clean, responsive Flask-based UI
• Local Inference: Run completely offline with local models
• Performance Monitoring: Real-time GPU/CPU usage tracking
• Multiple Voices: Support for various voices and speakers
• Audio Generation: High-quality WAV output
• Logging: Comprehensive generation logs with performance metrics

🛠️ Supported Models

• Kokoro: Fast, lightweight TTS optimized for local macOS deployment

📋 Prerequisites

• macOS (Intel or Apple Silicon)
• Python 3.10+
• Homebrew

🏗️ Installation

Automated Setup (Recommended)

Run the setup script to automatically configure your environment:

./setup-neo-tts.sh

This will:

• Install Homebrew dependencies (ffmpeg, pkg-config, Python 3.10)
• Create a virtual environment
• Install all required Python packages
• Download TTS models
• Set up project directories

Manual Setup

If you prefer manual installation:

1. Install system dependencies:

   brew install ffmpeg pkg-config python@3.10

2. Create virtual environment:

   python3.10 -m venv venv
   source venv/bin/activate

3. Install Python dependencies:

   pip install -r requirements-neo.txt

4. Download models:

   python -c "
   from huggingface_hub import snapshot_download
   snapshot_download(repo_id='hexgrad/Kokoro-82M', local_dir='models/kokoro_cache')
   "

🚀 Usage

1. Activate virtual environment:

   source venv/bin/activate

2. Start the server:

   python app/app.py

3. Open your browser: Visit http://localhost:5000

4. Generate speech:

   • Select a TTS model
   • Choose a voice
   • Enter your text
   • Click generate

📁 Project Structure

local-tts-devlopment/
├── app/                    # Flask application
│   ├── static/            # Static assets (CSS, JS, output files)
│   ├── templates/         # HTML templates
│   ├── app.py            # Main Flask application
│   ├── device_utils.py   # Device monitoring utilities
│   └── __init__.py
├── models/                # TTS model implementations
│   └── kokoro.py         # Kokoro TTS wrapper
├── logs/                  # Generation logs and metrics
├── venv/                  # Virtual environment (created by setup)
├── setup-neo-tts.sh      # Automated setup script
├── requirements-neo.txt  # Python dependencies
└── README.md             # This file

🔧 Configuration

The application automatically detects your hardware and optimizes accordingly:

• Apple Silicon (M1/M2/M3): Uses Metal acceleration when available
• Intel Macs: Falls back to CPU inference
• GPU Monitoring: Tracks VRAM usage and performance metrics

📊 Monitoring

The application provides real-time monitoring of:

• CPU/GPU usage during generation
• Generation time and audio duration
• Model performance metrics
• Device information

Access monitoring data via the /api/device-info endpoint.

🧪 Testing

Run the included verification tests:

# Test Kokoro model
python -c "
from models.kokoro import list_voices, generate_audio
voices = list_voices()
if voices:
    generate_audio('Hello from Kokoro', voices[0])
    print('✅ Kokoro OK')
"

🤝 Contributing

1. Fork the repository
2. Create a feature branch: git checkout -b feature-name
3. Make your changes
4. Test thoroughly
5. Submit a pull request

Adding New Models

To add support for new TTS models:

1. Create a new module in models/
2. Implement the required interface:
   • list_voices(): Return available voices
   • generate_audio(text, voice, output_path): Generate audio file
3. Register the model in app/app.py MODELS dictionary

📝 License

This project is licensed under the MIT License - see the LICENSE file for details.

🙏 Acknowledgments

• Kokoro TTS - Fast local TTS
• Flask - Web framework
• PyTorch - Machine learning framework

🐛 Issues

If you encounter any issues:

1. Check the logs in app/logs/results.csv
2. Ensure your virtual environment is activated
3. Verify all dependencies are installed
4. Check device compatibility

For bugs or feature requests, please open an issue on GitHub.