OuteTTS provides a unified Python interface for running OuteAI’s text-to-speech models. It supports multiple inference backends including llama.cpp Python bindings, llama.cpp server, Hugging Face Transformers, ExLlamaV2, and vLLM. The project offers both single inference and batched (async) modes for serving TTS models, with published model checkpoints on Hugging Face at 0.6B and 1B parameter sizes.