VTT is WebVTT, the only subtitle format a browser reads natively. SRT is the subtitle format everything reads, and the simplest one. ffmpeg sits between the two, and this page says exactly what happens to a VTT file on the way to becoming an SRT one.
What runs when a VTT file becomes SRT
ffmpeg held to a single thread, on a machine we rent and watch. Your VTT file is uploaded once, ffmpeg runs once, the SRT comes back, and neither file is kept. VTT to SRT is one of the 3649 pairs that engine was probed on with a real file, which is why it has a page here and why the pairs the probe could not prove do not.
What survives from the VTT into the SRT
Every line, and the moment it appears. The engine reads your VTT and writes the same cues into the SRT: the text is not re-wrapped, the order is not touched, and the start of each cue lands where it was. It counts the cues on the way out and refuses the result if the count moved.
Accents and every other character. The SRT is written as UTF-8, which is the one thing that stops subtitles arriving as question marks on a machine that is not yours.
A file whose size you can predict. We put the same 200-cue file through all six formats and weighed what came out: 88 bytes per cue in SRT, 12 % larger than the VTT, because that is a number, two timestamps, the text and a blank line, once per cue. The extra bytes buy something, and the next section says what.
End times as well as starts, exactly as your VTT had them. Nothing is recomputed.
What VTT to SRT costs you
Nothing measurable, which is worth saying plainly. VTT and SRT carry the same cues with the same fidelity, and this pair changes the container rather than the content.
Where VTT and SRT files come from
VTT. A W3C standard, and the only format the HTML track element accepts. If subtitles are going on a web page, this is not a preference but a requirement. It is SRT with a header line, dots instead of commas, and optional styling and positioning.
SRT. Written for SubRip, a Windows program that pulled subtitles off DVDs, and it won by being trivial: a number, two timestamps, the lines, a blank line. No styling, no specification, no committee. Every player on earth reads it.
If a model is going to read the SRT
A subtitle file is a transcript with the clock attached, which is what makes it worth feeding to a model at all: every line comes with the moment it was said. VTT in, SRT out, and the START of every line lands where it was, so a question like "what was said at eleven minutes" still has an answer afterwards.
VTT to SRT, measured rather than promised
Running VTT to SRT against the real engine with a real file, ffmpeg wrote 137 bytes of SRT in 0.1 seconds, machine otherwise idle. That is one VTT file on one day and not an average, which is why the number is given together with the file that produced it.
The VTT to SRT verdict was reached by reading the text, the timings and the length of every cue, meaning the output was read back and recognised. Which check was used matters, because they do not all prove the same thing, and a status code of 200 proves nothing whatsoever about whether the SRT file has anything inside it.
Other ways into SRT, and what they measured
Among the published routes into SRT, VTT is the second largest output of the 3 measured. The witness files differ, so this ranks the probe run and not your document.
ASS to SRT: 137 bytes in under a tenth of a second.
LRC to SRT: 137 bytes in 0.1 seconds.
What we will not pretend about VTT to SRT
A VTT file over 5 MB is refused before the upload finishes rather than after it, so you do not wait for a rejection.
A VTT to SRT run that passes 30 seconds is killed, and the ffmpeg process is killed with it. A run left behind would sit on one of the machine's two cores until somebody noticed.
This family does not touch the timing itself. Shifting subtitles by two seconds, stretching them for a different frame rate or merging two files are real needs and none of them happens here: this converts a container into another container.
One VTT file at a time, chosen in the browser. There is nothing else to set up and nothing else on offer.
VTT to SRT: what people ask
What actually converts my VTT file to SRT?
ffmpeg does it, held to a single thread, on a machine we rent and watch. Not a browser trick and not somebody else service: the VTT file is uploaded once, ffmpeg runs once, the SRT comes back, and neither file is kept afterwards.
How long does VTT to SRT take?
On the file the probe used, ffmpeg took 0.1 seconds and wrote 137 bytes of SRT. That is one real measurement on one real VTT file, not an average and not a promise about yours: a larger VTT takes longer, and past 30 seconds the run is stopped.
What do I lose going from VTT to SRT?
The one to know about first: Nothing measurable, which is worth saying plainly. VTT and SRT carry the same cues with the same fidelity, and this pair changes the container rather than the content.
Will the timings still match my video?
Yes. The start of every cue is carried across unchanged, not recomputed. The ends too, exactly as your file had them.
Is VTT to SRT free?
There is a free allowance every month, and one VTT to SRT conversion costs half a credit against it. When the allowance runs out the tool says so and stops, rather than quietly handing you a worse SRT.