RVC-Boss/GPT-SoVITS ? reverse-engineered prompt

Build me a simple voice cloning and text to speech web app like GPT SoVITS.

I want to upload a short voice sample and type text, then get speech back in that voice. It should also support training or improving the voice with about a minute of audio, so the result sounds more realistic. Please include an easy web interface for beginners, with tools to help prepare audio clips, split long recordings, separate vocals from background music if possible, transcribe Chinese audio, and label text for training.

It should work across languages where the project supports it, especially English, Chinese, Japanese, Korean, and Cantonese. I’d like both quick zero shot use with a few seconds of audio and a better few shot workflow with more data.

Set it up as a Python project that can run locally, with clear install and run scripts for Windows, Linux, macOS, and Docker if practical. Look up current docs online if you need to.