SamTTS - Multi-Backend Offline Text-to-Speech API

A unified HTTP REST API providing access to multiple offline Text-to-Speech (TTS) engines through a single interface. Built on top of 🐸Coqui TTS and other leading TTS technologies.

🚀 Features

• Unified API - Single HTTP interface for multiple TTS backends
• Offline Operation - No internet connectivity required
• Multiple Engines - Support for Coqui TTS, eSpeak, eSpeak-NG, MaryTTS, pyttsx3, and Festival
• Auto-Detection - Automatic detection of available TTS engines
• Streaming & Batch - Real-time streaming and batch synthesis support
• Voice & Language Support - Multiple voices and languages per backend
• Adjustable Parameters - Control speed, pitch, and other speech parameters

🎯 Quick Start

Start the API Server

bash cd multi_tts_api python -m uvicorn api:app --host 0.0.0.0 --port 8000

Synthesize Speech

bash curl -X POST http://localhost:8000/synthesize \ -H "Content-Type: application/json" \ -d '{"text": "Hello world", "backend": "espeak"}' \ --output speech.wav

📋 Supported TTS Backends

| Backend | Quality | Speed | Languages | Features | |---------|---------|-------|-----------|----------| | Coqui TTS | Excellent | Medium | 20+ | Neural models, voice cloning | | eSpeak | Fair | Very Fast | 50+ | Lightweight, reliable | | eSpeak-NG | Good | Very Fast | 100+ | Improved quality | | MaryTTS | Good | Medium | Multiple | Modular, customizable | | pyttsx3 | Varies | Fast | Varies | Cross-platform wrapper | | Festival | Good | Medium | Multiple | Highly configurable |

📖 API Documentation

Core Endpoints

List Available Backends

bash GET /backends

Get Backend Information

bash GET /backends/{backend_id}

Synthesize Speech

```bash POST /synthesize Content-Type: application/json

{ "text": "Hello, this is a test", "backend": "espeak", "language": "en", "speed": 1.0, "pitch": 1.0, "format": "wav" } ```

Streaming Synthesis

bash POST /synthesize/stream

Batch Synthesis

bash POST /synthesize/batch

For detailed API documentation, see multi_tts_api/README.md.

🛠️ Installation

Prerequisites

Install system dependencies for the TTS backends you want to use:

```bash

For eSpeak

sudo apt-get install espeak

For eSpeak-NG

sudo apt-get install espeak-ng

For Festival

sudo apt-get install festival

For MaryTTS (requires Java)

sudo apt-get install default-jre ```

Python Dependencies

```bash

Clone the repository

git clone [your-repo-url] cd SamTTS

Install Python dependencies

cd multittsapi pip install -r requirements.txt ```

🎭 Backend Details

Coqui TTS

Based on the original 🐸TTS library with support for: - High-performance Deep Learning models (Tacotron2, Glow-TTS, VITS, YourTTS) - Neural vocoders (HiFiGAN, MelGAN, WaveRNN) - Multi-speaker synthesis and voice cloning - 20+ languages with pretrained models

Other Backends

• eSpeak/eSpeak-NG: Lightweight, rule-based synthesis
• MaryTTS: Modular Java-based platform
• pyttsx3: Cross-platform TTS wrapper
• Festival: Configurable speech synthesis system

💻 Usage Examples

Python Client

```python import requests

Start the API server first

python -m uvicorn multittsapi.api:app --host 0.0.0.0 --port 8000

base_url = "http://localhost:8000"

List available backends

response = requests.get(f"{base_url}/backends") backends = response.json() print("Available backends:", backends)

Synthesize with eSpeak

tts_request = { "text": "Hello from SamTTS!", "backend": "espeak", "language": "en", "speed": 1.2 }

response = requests.post(f"{baseurl}/synthesize", json=ttsrequest) with open("output.wav", "wb") as f: f.write(response.content) ```

Command Line Interface

```bash

List available backends

curl http://localhost:8000/backends

Synthesize speech with different backends

curl -X POST http://localhost:8000/synthesize \ -H "Content-Type: application/json" \ -d '{"text": "Fast synthesis", "backend": "espeak"}' \ --output fast.wav

curl -X POST http://localhost:8000/synthesize \ -H "Content-Type: application/json" \ -d '{"text": "High quality synthesis", "backend": "coqui"}' \ --output quality.wav

Batch synthesis

curl -X POST http://localhost:8000/synthesize/batch \ -H "Content-Type: application/json" \ -d '[ {"text": "First sentence", "backend": "espeak"}, {"text": "Second sentence", "backend": "festival"} ]' \ --output batch.zip ```

📁 Project Structure

├── multi_tts_api/ # Main API application │ ├── api.py # FastAPI application and endpoints │ ├── backend_manager.py # Backend management and orchestration │ ├── backends/ # TTS backend implementations │ │ ├── base.py # Abstract base class for backends │ │ ├── coqui.py # Coqui TTS backend │ │ ├── espeak.py # eSpeak backend │ │ ├── espeak_ng.py # eSpeak-NG backend │ │ ├── festival.py # Festival backend │ │ ├── marytts.py # MaryTTS backend │ │ └── pyttsx3.py # pyttsx3 backend │ ├── requirements.txt # Python dependencies │ ├── run_server.py # Server startup script │ ├── test_api.py # API test suite │ └── README.md # Detailed API documentation ├── TTS/ # Original Coqui TTS library └── README.md # This file

🤝 Contributing

Contributions are welcome! Areas for improvement: - New TTS backend implementations - Performance optimizations - Additional audio format support - Better error handling and logging - Extended language support

📄 License

This project builds upon multiple open-source TTS libraries: - Coqui TTS: Mozilla Public License 2.0 - eSpeak/eSpeak-NG: GPL v3
- MaryTTS: LGPL v3 - Festival: Custom license

See individual backend documentation for specific license requirements.