neonwatty/bleep-that-shit — reverse-engineered prompt

Build me a private browser app for censoring audio and video clips. I want users to upload an MP3 or MP4, have the app transcribe it in the browser, then choose words or phrases to bleep out. Nothing should upload to a server, the media and transcript should stay on the user’s device.

The main flow should feel simple, upload a file, pick a language and transcription model, run transcription, show the detected words with timestamps, let the user enter profanity or custom words to censor, support exact, partial, and fuzzy matching, then preview and download the censored file. Include a few bleep sound choices.

Also make a simple model testing page where users can compare transcription speed and accuracy for different Whisper model sizes, like tiny, base, and small. Keep it friendly and clear for non technical creators who just want to make clips more ad friendly or make funny bleeped videos. Use in browser speech to text and media processing, and look up current docs online if needed.