coqui-ai/TTS ? reverse-engineered prompt

Build me an open source Python text to speech toolkit that can turn written text into realistic speech, using ready made models as well as custom trained ones.

I want users to be able to install it, pick a voice model, type some text, and save the spoken audio to a file. It should support many languages, different voices, voice cloning from a short sample when the model allows it, and voice conversion if that is supported. Please include simple command line usage and Python examples so someone can try it quickly without reading a research paper.

Also add tools for people who want to train or fine tune voices, with clear logs, dataset checking helpers, and example recipes they can copy. Keep it modular so researchers can add new speech models, vocoders, and speaker encoders later.

Please include documentation, tests, notebooks or examples, and Docker support so it works for both experiments and production use. Look up current docs online if you need to.