OpenBMB/VoxCPM ? reverse-engineered prompt

Build me a web application for text-to-speech using the VoxCPM2 model. I want a simple interface where I can type or paste text and generate realistic, high-quality audio.

The app needs to handle multiple languages automatically—I should be able to input English, Spanish, Japanese, Chinese, and others without having to manually select the language.

The coolest feature I want is voice cloning. The user should be able to upload a short audio file (like a 10-second voice clip), and the app will clone that voice to read any new text. For these cloned voices, I'd also like the ability to add simple instructions to control the style, like "(speak a little faster)" or "(sound more cheerful)".

Also, please add a "voice design" feature where I can just describe a voice in text—like "(A young woman with a gentle, sweet voice)"—and it will create a brand new voice from that description. Use Python for the backend and something simple like Gradio or Streamlit for the web UI.