On YouTube, I change the speed constantly, and, sometimes, I think I'm watch at x1 when I'm actually watching at x1.5 or even x2.
I can usually differentiate x1 from other speeds, but when there's a slow talker and no music, it's actually quite easy not to be able to tell unless I think hard about it, or realistically just check, especially since I'm very used to it, having done that for around two years.
However, watching a show for 30 20 minutes without realizing it is unrealistic, especially since there are a lot of sound effects and it's 30 20 minutes!
The ChatGPT app uses the speech-to-text model Whisper, and it's always spot on in English. Whisper is open source. I don't understand why it's not widespread, but hopefully it will be or similarly capable software will be soon.
Does Lemmy not support alt text? I can't add alt text from Infinity for Lemmy; is that because Lemmy doesn't support it or just my client? Because they really should.
Fun fact: Most receipts don't use ink. Just thermal paper and heat.