Fatterbox TTS server

A lightweight, production-ready TTS server that supports OpenAI-compatible API and Wyoming Protocol for voice synthesis. Built for easy deployment with Docker and minimal configuration.

---

🐳 Docker Image

Use the pre-built Docker image from Docker Hub:

docker pull docker.io/justinlime/fatterbox:latest

This image includes all dependencies, optimized for GPU/CPU inference and ready to run.

---

🚀 Quick Start

Run the Container

docker run -p 5002:5002 -p 10200:10200 \
  -e HOST=0.0.0.0 \
  -e OPENAI_PORT=5002 \
  -e WYOMING_PORT=10200 \
  -e DEVICE=cuda \
  -e AUDIO_PROMPT=/audio/prompt.wav \
  -e LANGUAGE_ID=en \
  -e STREAM=true \
  -e EXAGGERATION=0.5 \
  -e TEMPERATURE=0.8 \
  -e MIN_P=0.05 \
  -e TOP_P=1.0 \
  -e REPETITION_PENALTY=1.2 \
  -e DTYPE=bfloat16 \
  -v /path/to/audio/prompt.wav:/audio/prompt.wav \
  --name chatterbox-tts \
  docker.io/justinlime/fatterbox:latest

Replace /path/to/audio/prompt.wav with the actual path to your audio prompt file.

---

🛠️ Environment Variables

Variable	Description	Default
HOST	Server IP	0.0.0.0
OPENAI_PORT	OpenAI API port	5002
WYOMING_PORT	Wyoming protocol port	10200
DEBUG	Enable debug logs	false
DEVICE	cpu, cuda, or mps	cpu
STREAM	Enable streaming	true
LANGUAGE_ID	Language ID	en
AUDIO_PROMPT	Path to audio prompt	(required)
EXAGGERATION	Voice expression	0.5
TEMPERATURE	Sampling temperature	0.8
MIN_P	Nucleus sampling	0.05
TOP_P	Top-p sampling	1.0
REPETITION_PENALTY	Repetition control	1.2
DTYPE	Model precision	bfloat16

---

🚀 Usage

OpenAI-Compatible API

curl -X POST http://localhost:5002/v1/audio/speech \
  -H "Content-Type: application/json" \
  -d '{
    "input": "Hello, this is a test.",
    "voice": "alloy",
    "response_format": "mp3"
  }'

Wyoming Protocol

Connect via:

wyoming-client --server tcp://localhost:10200 --text "Hello world"

---

🧩 Features

• ✅ OpenAI API compatibility
• ✅ Wyoming Protocol support
• ✅ Voice cloning with audio prompt
• ✅ Multi-lingual support
• ✅ GPU acceleration (CUDA/MPS)
• ✅ Mixed precision (bfloat16/float16)
• ✅ Streaming & batch inference
• ✅ Low-latency design

---

🧩 Model Requirements

• Chatterbox TTS model (pretrained)
• Audio prompt file (for voice cloning)
• Supported languages: en, de, fr, es, etc.

---

🧪 Testing

Test API

curl -X POST http://localhost:5002/v1/audio/speech \
  -H "Content-Type: application/json" \
  -d '{
    "input": "The quick brown fox jumps over the lazy dog.",
    "voice": "nova",
    "response_format": "mp3"
  }' > output.mp3

Test Wyoming

wyoming-client --server tcp://localhost:10200 --text "Hello world"

---

🧩 Build & Run

Use Pre-Built Image

docker pull docker.io/justinlime/fatterbox:latest
docker run -p 5002:5002 -p 10200:10200 -e DEVICE=cuda -v /path/to/audio/prompt.wav:/audio/prompt.wav docker.io/justinlime/fatterbox:latest