babysor/MockingBird ? reverse-engineered prompt

Build me a Python app that can clone a person’s voice from a short audio sample and then generate new speech from typed text. I want it to focus on Mandarin Chinese support, but it should still feel like a general voice cloning demo.

Please include a simple way to run it locally, with either a desktop demo or web interface where I can upload or choose a voice sample, type some text, and hear or download the generated audio. It should also support using pretrained models when available, and have scripts for preprocessing data and training a synthesizer if someone wants to improve the voice quality with their own dataset.

Make the setup clear for Windows and Linux, and keep GPU support in mind because this is deep learning audio work. If needed, look up current docs online so the install instructions don’t rely on old package versions.