From what I remember TheFluff was arguably one of the best timers (and encoders) in the fansub scene in the past and he could time the dialogue in under 10 minutes. https://www.youtube.com/watch?v=00xX0PwGUg8

Karaoke and typesetting can of course take longer (I remember someone complaining about how much effort it was to typeset every single book name in some scene that had a bookshelf) though karaoke is usually ~3 minutes of unique content (OP & ED) per ~12 episodes. Typesetting depends heavily on anime, like isekais don't usually have a lot Japanese writing anyway.