francozanardi/tscaps — reverse-engineered prompt

Build me a browser based video caption editor like the open source local version of tscaps. I want people to drop in a video, transcribe speech right in the browser with Whisper, edit the captions, choose from a gallery of caption looks, and preview those captions live as styled HTML over the video. Then they should be able to export a brand new video with the captions burned in, all on the device, with no uploads, no account, and no backend.

Please include the reusable rendering engine as well as a simple web app that wraps it. The caption looks should come from template folders with JSON and CSS so adding a new style is basically just dropping in a folder. Keep the whole thing easy to run locally, easy to self host, and able to build as a static site or Docker image. If anything is unclear, check the repo docs and current browser docs online and make sensible choices.