EPUB TTS Reader API

Testing phase (All features might not work)

A REST API for converting EPUB files to text and generating text-to-speech audio using the Supertonic-3 model. Supports direct text input, EPUB conversion via Calibre, and sentence-level audio generation with on-device inference.

Technology Stack

Backend

• FastAPI: Modern, fast web framework for building APIs with Python
• Supertonic-3: Lightning-fast, on-device text-to-speech model supporting 31 languages
• ONNX Runtime: High-performance inference engine for machine learning models
• Calibre: eBook conversion tool for EPUB to text conversion
• Python 3.12: Core programming language

Frontend

• HTML5/CSS3: Markup and styling
• Vanilla JavaScript: Client-side interactivity
• Flexbox: CSS layout system

Key Features

• On-device TTS inference (no cloud API calls required)
• EPUB to text conversion via Calibre CLI
• Sentence-level audio generation
• Click-to-play from any text position
• Auto-scroll toggle functionality
• Adjustable font size
• Monospace font with borders for clarity
• RESTful API with OpenAPI/Swagger documentation

API Documentation

Interactive API documentation is available at:

• Swagger UI: http://localhost:8000/docs
• ReDoc: http://localhost:8000/redoc

Installation

Local Development

Prerequisites

• Python 3.12+
• Calibre (for EPUB conversion)
• uv (Python package manager)

Setup

1. Clone the repository and navigate to the project directory

2. Create a virtual environment and install dependencies:

uv venv
uv pip install fastapi uvicorn python-multipart supertonic

3. Set your Hugging Face token as environment variable or update in main.py

4. Run the server:

.venv\Scripts\python main.py

The API will be available at http://localhost:8000

Docker Deployment

Prerequisites

• Docker installed on your system

Build and Run

1. Build the Docker image:

docker build -t epub-tts-reader .

2. Run the container:

docker run -d -p 8000:8000 -e HF_TOKEN=your_huggingface_token epub-tts-reader

3. Access the API at http://localhost:8000

Docker Compose

Create a docker-compose.yml file:

version: '3.8'
services:
  epub-tts-reader:
    build: .
    ports:
      - "8000:8000"
    environment:
      - HF_TOKEN=your_huggingface_token
    volumes:
      - ./uploads:/app/uploads
      - ./audio:/app/audio

Run with:

docker-compose up -d

API Endpoints

Web Interface

• GET / - Main web interface for EPUB TTS Reader

EPUB Processing

• POST /upload-epub - Upload and convert EPUB to text

Text Processing

• POST /load-text - Load and parse direct text input

TTS Generation

• POST /generate-audio - Generate audio for text segment

Audio Serving

• GET /audio/{filename} - Serve generated audio files

Deployment Considerations

Security

• Remove or secure the Hugging Face token before public deployment
• Implement rate limiting for API endpoints
• Add authentication/authorization for production use
• Validate and sanitize all user inputs
• Use environment variables for sensitive configuration

Performance

• Implement audio file cleanup to prevent disk space issues
• Add caching for frequently accessed content
• Consider using a CDN for static assets
• Implement request queuing for heavy TTS operations

Scalability

• Use a production ASGI server like Gunicorn with Uvicorn workers
• Implement horizontal scaling with load balancing
• Use Redis or similar for session management
• Consider containerization with Docker

Monitoring

• Add logging for API requests and errors
• Implement health check endpoints
• Set up error tracking (e.g., Sentry)
• Monitor resource usage and response times

Environment Variables

• HF_TOKEN: Hugging Face API token for model access
• PORT: Server port (default: 8000)

License

MIT