Speech recognition is good on clear speech and weak on names, numbers, jargon, accents, music and people talking over each other. A wrong word in a caption is easy to miss unless you read for it.
If every caption is a little early or late, shift all times by the same amount (for example 0.4 seconds) instead of editing each one.
The Caption Maker has all of this next to the video: click a caption to jump to it, edit text and times, find and replace, split, join and shift.
Open the free Caption MakerLonger files and offline launch files are planned for a one-time $24 pack, not yet for sale: Pack in preparation