As much as hate generative AI in artistic fields with a burning passion, it sounds like you have some serious emotional instability if this rocks your world to the point that you fall into persistent depression and burn bridges with all your friends. I’m very sorry you feel this way and hope for everyone’s sake that human-made music stays appreciated going forward.

Not really accurate seeing as voting rates among the poor and uneducated are substantially lower, even though they are the ones most often impacted negatively.

I usually just find a short isolated piece of dialogue (or the first spoken word) and count the time between the audio and the subs, whichever comes first. That usually gets me close to +/- 1s. Then I micro adjust from there.

You just gotta remember that a negative offset makes the subs appear earlier, and vice versa for a positive offset.

Except when they are shit and contain stuff such as #everyone_dies and #main_character_dies, like the default metadata source for Jellyfin and some torrent aggregators do.