Sadly, doesn't work too great in this situation:
> That they didnt go through but i would tell you theyre just a chill look at here lets do it chills with all of our great men and they look at every chance they go oh do you want to the black man well thats my gosh thats my gosh thats my gosh thats my gosh thats my gosh thats my gosh thats
Uploaded video dialog:
Bowman: You know, of course, though, he's right about the 9000 series having a perfect operational record. They do.
Poole: Unfortunately that sounds a little like famous last words.
Bowman: Yeah, still it was his idea to carry out the failure mode analysis, wasn't it?
Poole: mmm
Bowman: Should certainly indicate... (away from camera): his integrity and self-confidence
Bowman: If he were wrong, it'd be the surest way of proving it.
Poole: It would be if he knew he was wrong.
Results:
"Of course there is recommended getting necessary to have a perfect operational rank i know youre going to be the first to do that youre going to get the best youre going to get the best youre going to get the best youre going to get the best youre going to get the best of yours if you want to rock better sure its well perfect."
I'd be interested to know how accurate it is, from what angles it will read lips at (front facing, side, etc).
Sounds promising if it works well. Imagine all the historical videos without sound you could try to finally know what was being said.
'i love you' and 'island view' have the same lip movements is the clasical example.
my mother was a mostly-deaf lip-reader. She needed conversational context in order to keep up 'legibly'; and it created a lot of fun between the two of us when she would come up with an oddball question or comment that had nothing to do with the conversation once-in-awhile when her guesses failed spectacularly.
With context, though, it's a great tool. She and I used to watch crime dramas with the sound off late at night and never miss a beat. It feels if you're trying to transcribe something that has a lot of structural context the success rate is higher than 50%, but I don't know that formally.
It's still a tool I use in conversation. Even with good hearing it's tough to hear people in crowded restaurants or concert venues, lip-reading helps immensely.
User-submitted videos (with audio for STT), user-crafted bounding boxes (we might not need these soon), and user-guided RLHF.
The submitted videos are likely diverse, challenging (otherwise the human might just do it), and representative of solving actual customer problems.
