What an SRT cannot carry
A WebVTT can bring along things an SRT has no place for, and what matters is what happens to them rather than letting them through. The positioning settings after the timestamp — line, align, position — can't be translated into anything, so they're dropped. NOTE and STYLE blocks aren't subtitles and mustn't end up inside the text either, so they go.
Voice tags are the interesting case. In a VTT the speaker is marked with a tag around the line; in an SRT that doesn't exist. Rather than throw it away — which would lose who says what — it becomes the name in front of the line, followed by a colon, which is how dialogue has always been written in a subtitle.