SRT vs VTT vs TXT: which subtitle format do you need?
Updated 15 November 2026
The short answer
- Uploading to YouTube, Vimeo, Premiere, DaVinci or a video editor — SRT. It is the format every one of them reads without argument.
- Embedding subtitles in a web page with <video> — VTT. It is what browsers implement.
- Reading, searching, quoting, feeding to another tool — plain text. Timestamps are noise when you are reading.
What SRT looks like
1
00:00:01,000 --> 00:00:04,200
This is the first line.
2
00:00:04,400 --> 00:00:07,900
And this is the second.
Three parts per cue: a sequence number, a timestamp range, and the text. Nothing else is required. That simplicity is exactly why it works everywhere — it predates the web and every tool that handles subtitles handles this.
The timestamp is HH:MM:SS,mmm — hours, minutes, seconds, and milliseconds separated by a comma. Remember the comma.
What VTT looks like
WEBVTT
00:00:01.000 --> 00:00:04.200
This is the first line.
00:00:04.400 --> 00:00:07.900 align:middle line:90%
And this is the second.
WebVTT is the W3C format for HTML5 <video>. It requires the WEBVTT header line, uses a period before the milliseconds instead of a comma, and drops the sequence numbers.
What it adds over SRT: cue settings (position, alignment, line), CSS styling of the cue, and regions. If your subtitles need to sit in a particular corner or take a particular font, that is VTT, not SRT.
What it does not add: universal editor support. Some older editors accept VTT but quietly mangle the styling.
The comma-versus-period trap
This is the single most common cause of "the subtitles do not load" and "the subtitles are off by a frame":
- SRT:
00:00:01,000— comma - VTT:
00:00:01.000— period
A VTT file with commas in its timestamps is invalid. An SRT file with periods may load in lenient players and fail in strict ones. If you renamed .vtt to .srt without converting the timestamps, that is the bug.
Plain text
A .txt transcript has no timing information. For a meeting record, an article draft, a quote you are pulling, or anything that will be read rather than watched, that is the correct export. Adding timestamps to a document nobody will play along to just makes it harder to read — which is why most transcript exports offer both versions.
If you need to locate a specific moment later, timestamps do help. A common middle ground is a plain-text transcript with a timestamp every few paragraphs, rather than one per line.
What about the other formats?
You will also run into ASS and SSA (styled subtitle formats used by anime fansub tools), SBV (YouTube's older export), and TTML (broadcast). If you have one of those and need SRT, converting is a format transform, not a re-transcription — the text and timing stay the same, only the container changes.
Practical rule
Export SRT unless something specifically demands otherwise. It is the format least likely to be rejected, and every other format can be produced from it later. Change the format when you know the destination; do not guess.
Already have subtitles inside another container? Extract the subtitle track from an MKV →
Have the text but no timings? Turn plain text into an SRT file →
Need to convert one to the other? WebVTT to SRT converter →
Other formats: SBV · SAMI · MicroDVD · SRT to plain text