It'll never happen. They refuse to pay people to do that work, and we'll be stuck with hallucinated dialog for decades.
me, ears failing: *watches tv* character on screen: "tomato-borne civil soup" me: "wtf... lemme turn on subtitles" AI-generated subtitles: "tomato born in civil soup" me: *looking forward to the future, relayed to me by an hallucinating e-tween who's worse at accents than Arnold Schwarzenegger*
@penryu @wreimers To be sure, too many videos have unedited and messy captions. No disagreement there.
That said, there are people who are cleaning up captions and transcripts as part of their job duties, and getting paid for it. Whether it's happening enough is debatable, as is whether they are getting paid enough, but the editing and cleaning up? That is not something that will never happen. It's happening now. And there are people being trained on how to do this properly. Thankfully!
I used to see great captions, only the occasional "[unintelligible]" or outright wrong text, where the living person just couldn't make it out. And in some cases I could tell the person listening just didn't have much experience with British dialog — fair enough.
Lately they're getting worse throughout, where it's clear there's an algorithm just finding a closest fit, without even bothering to check if it actually fits.
So, I do appreciate any effort to improving it!