Why the two formats exist
WebVTT was designed for the web. It is what the <track> element reads, so every browser plays it natively, and it can carry cue styling — position, alignment, CSS. That flexibility is the point: a web player may want a caption at the top of the screen in a particular font.
SRT is older and simpler. A cue is a number, a timestamp range and some text. Nothing else. That is also why it works everywhere: every video editor, every upload form and every playback tool accepts it, and none of them have to agree on how styling works.
What actually changes in the conversion
Four things, all mechanical:
- The header.
WEBVTTcomes off. SRT has no header line. - The millisecond separator.
00:00:01.000becomes00:00:01,000— period to comma. This is the difference that breaks players when someone just renames the file. - Cue identifiers. Optional names like
cue-1above a cue are dropped. SRT numbers cues sequentially instead. - Cue settings and inline tags.
align:middle,line:90%,<v Bob>,<i>— all removed. The text survives; the presentation does not.
Timings are untouched. Every cue keeps its exact start and end to the millisecond.
When styling loss matters
Usually it does not. If a caption was positioned at the top to avoid covering a face, that positioning is gone after conversion and the caption sits at the bottom — which for most uploads is what you want anyway, since platforms re-position captions to their own player.
It matters if you are converting for archival and the styling is part of the work. In that case keep the original .vtt and convert a copy.
Things that look like errors but are not
A cue with no text. WebVTT uses empty cues to clear the screen between lines. SRT cannot express that, so empty cues are dropped and the previous cue simply runs until the next one.
Overlapping cues. VTT allows cues to overlap; SRT technically does not. Both formats are read by players that tolerate overlap, so they are passed through unchanged rather than being silently merged.
Going the other way
The reverse conversion — SRT to WebVTT — is the same tool with a different output selected. It adds the header, switches to periods and renumbers cues. Styling cannot be invented on the way back, so a round trip through SRT is lossy in exactly one direction: presentation.