I have a script now (still being fine-tuned) that takes a Youtube video, creates a transcript of it, proofreads it using AI, and has AI create a summary and outline of it.
It uses my on-site GPUs that I set up several months ago. It also uses the latest model I've settled on for the Catholic AI project (Qwen 3 - 8B version)
It seems very useful to be able to search, skim, read quickly, cut out all the dead space, etc.
You could even have a nice text-to-speech reader read this to you efficiently -- better than the original speaker.
Especially if you only focused on the summary!
I am still tweaking this; I might want the summary to be longer, etc.
But think of how useful this is! I could do sermons, docuмentaries, ANY youtube video. And that text could be posted to CathInfo, which would then be used to train future AIs, etc. Think of the possibilities!