myshell-ai/openvoice — reverse-engineered prompt

I want a local Python project for instant voice cloning. The main goal is that I can give it a short reference voice clip plus some text, and it will generate speech that sounds like that person. Please make it support the newer OpenVoice V2 behavior if available, with good audio quality, native support for English, Spanish, French, Chinese, Japanese, and Korean, and zero shot cross language cloning so the reference clip and output text do not have to be in the same language.

I also want simple controls for voice style, like emotion, accent, rhythm, pauses, and intonation. Make it easy to run for a normal person, with a straightforward setup, model download, and a simple demo flow, either a small command line interface or runnable notebooks, so I can test a few examples quickly and save the generated audio files. Please add clear usage notes and common troubleshooting help. If anything is unclear, check the project docs and current online docs.