The slow part isn't the text, it's the timings
Typing out what a video says is slow; giving every sentence the exact second it starts and ends is what turns an afternoon into a day. A ten-minute video is easily two hundred subtitles, each with two timestamps down to the millisecond.
That's what comes out of here already done. The timings are calculated against the audio of the file itself, so they line up as soon as you attach the SRT to the video — no drift, none of those global offsets you end up dragging around in the editor.