This project provides a complete inference server for the Chatterbox text-to-speech family, including the original, multilingual, and Turbo variants. It wraps the PyTorch-based TTS models behind a FastAPI service with an OpenAI-compatible endpoint, a Web UI, voice cloning support, and intelligent chunking for audiobook-scale text processing. The server supports GPU acceleration via NVIDIA CUDA, AMD ROCm, and Apple MPS, with CPU fallback.