Which should you choose: TXT, SRT or VTT?
Choose TXT if you want readable, exportable text: notes, quotes, edits, search, or something to paste into a document. Choose SRT if you need simple timed captions and the place you are uploading to accepts it. Choose VTT if the captions play on a web page through the HTML5 <track> element, if your destination recommends WebVTT, or if you need positioning, styling or chapters.
If you are not sure, export two: TXT for people to read and one caption format for the video. Both come from the same transcript, so nothing is lost by taking both. The table below sets the three formats side by side.
| TXT | SRT | VTT (WebVTT) | |
|---|---|---|---|
| Typical extension | .txt | .srt | .vtt |
| Timestamps | None, unless you add them | Start and end time on every cue | Start and end time on every cue |
| Structure | Plain lines or paragraphs | Numbered cues with a blank line between them | WEBVTT header first, then cues with optional identifiers |
| Timecode style | Not applicable | Comma before the milliseconds: 00:00:01,000 | Full stop before the milliseconds: 00:00:01.000 |
| Styling and positioning | None | None in the format | Cue settings and CSS styling |
| Easy for a person to read | Best: just the words | Awkward: numbers and timecodes between the lines | Awkward: header, timecodes and cue settings |
| Use as video captions | No: it has no timing | Yes, widely accepted | Yes |
| HTML5 <track> element | No | Not a defined format | Yes: the format it reads |
| Use as a transcript | Yes: reading, editing, search and quoting | Only after the timing is stripped | Only after the timing is stripped |
| Standard | None: plain text, so choose UTF-8 | None formally: a long-standing de facto format | A W3C specification (Candidate Recommendation Draft, 20 May 2026) |
What is a TXT transcript, and what does it leave out?
A TXT file is the words and nothing else: plain text that opens in any editor, pastes into any document, and can be searched, copied and quoted. That is why it is the right format for reading and editing a transcript. What it leaves out is timing. With no timestamps, a TXT file cannot drive captions, and you cannot turn it back into SRT or VTT without re-timing it, so keep the timed file if you may need captions later.
If you already have an SRT or VTT file and want the words, the SRT to TXT and VTT to TXT converters strip the cue numbers and timecodes in your browser. They can also keep each cue's start time in front of its line if you want a rough timed transcript.
SRT and WebVTT both carry timed captions
Both formats do the same core job: pair lines of text with start and end timecodes so a player shows the right caption at the right moment. The difference is lineage. SubRip's SRT is the older, plainer format – Matroska's technical reference calls it perhaps the most basic of all subtitle formats. WebVTT is the web-native one, published by the W3C Timed Text Working Group on the Recommendation track. No standards body governs SRT; it grew as a de-facto convention, which is why its rules stay loose.
You can tell them apart at a glance. A WebVTT file must open with the literal string WEBVTT on its first line, after an optional byte-order mark. An SRT file has no header at all. It starts straight into the first cue, numbered with a sequential index. That index above every SRT block is required; WebVTT makes the cue identifier optional.
SRT vs VTT: the comma-versus-full-stop timecode divide
The most cited difference is a single punctuation mark. SRT writes its milliseconds after a comma – the Matroska reference shows a cue running 00:02:17,440 --> 00:02:20,375. WebVTT writes the same field after a full stop, as in 00:02:17.440. The WebVTT grammar is explicit: a timestamp ends with a full stop followed by three ASCII digits for the thousandths of a second. Both formats use HH:MM:SS and a three-digit fraction; only the separator changes.
That one character is why the two files aren't interchangeable by a rename. A valid WebVTT file also needs its WEBVTT header, which an SRT file never carries. A converter has to add that header and swap every comma for a full stop, which is what the SRT to VTT converter does in your browser. If you only want the words, you can convert an SRT file to TXT instead. If you're building a caption file by hand, start from the plainer SRT structure and adjust from there.
Styling and positioning set the two apart
WebVTT carries formatting that SRT has no mechanism for. The spec defines six cue settings – vertical, line, position, size, alignment, and region – that place a caption anywhere in the frame. SRT has none of these. Matroska notes there are no general settings for SRT at all, so the format leaves that container field blank.
Styling goes further in WebVTT. CSS style sheets can target cues through the ::cue pseudo-element, and a file can hold STYLE blocks for reusable rules and REGION definitions for scrollable caption areas. SRT expresses none of that in the format. Its plainness is the point: if you need a caption parked in a specific corner or styled to match a brand, that lives in WebVTT.
Which format does your destination expect?
Your player decides the format more than your preference does. The HTML5 <track> element is defined to read WebVTT: MDN states the tracks are formatted in WebVTT format (.vtt files), and the WHATWG HTML standard ties the track element's URL to a WebVTT resource. The <track> element is specified only for WebVTT; SRT is not a defined <track> format.
WebVTT also carries more than captions. Through the <track> kind attribute, it can hold chapters, metadata, and descriptions. The WebVTT spec lists captions, subtitles, chapters, audio descriptions, and metadata as data kinds a file can carry. SRT defines no such mechanism. If your target is a web page, export a WebVTT file and point a <track> at it.
SRT's reach comes from its plainness. As the most basic subtitle format, it is the low-common-denominator file that many editors and upload fields accept, which makes it a sensible first try. It is not accepted everywhere, though: Vimeo recommends WebVTT, and PowerPoint reads SRT only from version 2411 on Windows. When you are unsure, check the destination's help page or export both. Exporting from a recording for those uploads? Generate an SRT instead.
| Destination | Accepts | Notes |
|---|---|---|
| YouTube | SRT and WebVTT, among others | SRT and SubViewer files must be plain UTF-8 |
| Vimeo | SRT and WebVTT | Recommends WebVTT; files must be UTF-8 |
| PowerPoint | WebVTT and SRT | SRT from version 2411 on Windows and 16.91 on Mac; SRT files must be UTF-8 |
| A web page | WebVTT | The HTML5 <track> element reads WebVTT only; serve the file as text/vtt |
Converting between TXT, SRT and VTT
Going from a timed file to plain text is easy; going back is not. Stripping the timing from a caption file loses information that a plain text file cannot hold, so keep the original SRT or VTT. The converters below all run in your browser, so the file is not uploaded.
SRT to TXT and VTT to TXT drop the cue numbers and timecodes and give you the words. SRT to VTT and VTT to SRT switch between the two caption formats, and the subtitle converter handles the rest, including detecting the format from the file.
Need a transcript or captions from audio or video first? Pepys Transcribe turns a recording into text and exports TXT, SRT and VTT, with your first minutes free.
Getting your captions onto the video
With the format settled, the next step is producing the file. Upload your recording and export it in the format you picked. From there, the file has to reach your player, and how you deliver it depends on where it's going.
On a web page, you host the .vtt and point an HTML5 <track> at it. A desktop player is different: side-load the .srt next to the video so the player finds it, or embed it in the file. Social platforms take the caption file in an upload field on the post. Getting the captions to display is its own step, separate from choosing the format, but the file you just exported decides how it runs.

