Minimal subtitle leading/lagging difference

Question

Lets start with a theory. Lets say you create subtitles perfectly synchronized with an audio track. Yes, I dont care about hearing impaired people atm, so video track doesnt concern me right now. Then you start delaying audio or subtitles. At what point will the listener notice the leading/lagging?

My goal is to create a minimal value on a scale used to time subtitles. So the user cant time subtitles beyond the point where it no longer has a meaning.

There must be a ton of research on this subject because all we hear is lagging behind what we see.

I just cant find the right keywords to google it.

Answer 1

There are lots of tricks to proper subtitling, but none of them involve delaying them. The reason you see large lags is because of "live" captioning -- they're appearing as fast as some trained operator can hear and enter them, which for the very best of them is still a few seconds.

For post-event titling, you want the text to appear at or slightly before the spoken word, but consistent with scene changes -- when a scene changes behind a title, the viewer tends to start reading it again, which is confusing.