Link: https://github.com/leestow/Censorarr

Censorarr is a self-hosted app that automatically creates profanity-filtered CLEAN audio tracks for movies and TV shows while preserving the original audio. It uses Whisper for speech transcription, can optionally use subtitles to improve detection, and integrates with Plex, Radarr, Sonarr, and Bazarr. It runs in Docker or natively on Windows, with optional remote GPU acceleration.

Well that just sounds terrible. I like my profanity.

Me too. But I got 2 little ones that like to repeat everything they hear.

Then don't let them watch it?

Or I can create this project and cut it out if I want.

This has been tried before. There were companies that would buy VHS (that's how old this was) and DVD movies, modify them to be clean, destroy the originals, replace them with a burned copy or duplicate, and sell them at markup. So families could buy "clean" versions of movies.

They were shut down because what they were doing was illegal, and rightly so.

I like your approach because it's done locally, which in and of itself, shouldn't be illegal. As I understand it, you're just muting the audio when bad words appear. You're just automating it.

Correct. All done locally on your server. It uses a small local AI model to detect the words and mutes them.