kigner/audio.cpp-webui ? reverse-engineered prompt

Build me a local audio app with a clean web interface that can run all the main audio tasks from one place.

I want to be able to type text and generate speech, upload or record audio and get transcriptions, do voice conversion and speaker detection, and also handle voice activity detection and basic audio cleanup. It should feel like a real desktop friendly tool, not a demo, with simple screens for choosing a model, downloading it, loading it, and running jobs. Please make it work well offline and keep everything local, since the whole point is avoiding Python dependency pain.

If there are multiple ways to run it, make a normal web app entry and a simple local launcher too. Use sensible defaults, show progress while models are downloading or tasks are running, and make the output easy to listen to, copy, or save. If you need to check current model or build docs online, do that.