SRT vs VTT vs TXT vs DOCX: Which File Do You Need?

Formats, platforms and Kukarella's exports checked on 18 October 2026.

Choose the file by where it is going. A video site, a social network or a video editor wants SRT. A video on your own web page, Moodle, Rise 360, or captions that must show in Canvas on an iPhone want VTT. People who will read, edit or print the words want DOCX (or a PDF made from it). An AI tool, a search or a quick paste wants TXT. But only SRT and VTT carry the timings that make words appear on screen in step with the speaker, so a Word transcript never counts as captions for a video, however accurate it is.

That last point is the one universities have been sued over. In the words of the W3C's media accessibility notes: "The full transcript supports different user needs and is not a replacement for captioning" (W3C, 2015). A transcript is the right answer for a podcast; a recorded lecture needs a caption file as well.

We make one of the tools in this story. Kukarella's TranscribeHub turns audio and video into text and exports Doc, Pdf, Txt and Srt. It does not export VTT, and in some cases its SRT gets the times wrong after the first hour. Both are covered below, with the converter we built to get round them. It runs in your browser and reads SRT and VTT files from any tool, but it cannot add timings to text that has none.

In this guide:

1. Which file each destination takes, in one table.
2. What actually differs between SRT and VTT.
3. A caption file or a transcript: what the accessibility rules ask for.
4. Turning a Zoom or Teams transcript into a Word document, and Word text into captions.
5. Why caption files fail to upload, and how long a caption line should be.

Which file does each platform take?

Short answer: look up the destination, not the format. Most video platforms and course tools take both SRT and VTT; a handful take only one. But every list below is what the platform's own help page said on 18 October 2026, and these lists change. Check the linked page if an upload fails.

The sources, platform by platform: YouTube and its untimed transcripts; Google Drive; Vimeo; Facebook; LinkedIn; X; the HTML track element; Canvas; Moodle; Rise 360 and Storyline 360; Captivate; Kaltura; Panopto; Premiere Pro; Final Cut Pro; Netflix; the BBC.

Two rows surprise people. Your own web page takes VTT only: the browser's caption track, the <track> element, is "formatted in WebVTT format (.vtt files)", and there is no SRT option (MDN). And broadcast does not run on SRT. The BBC asks for EBU-TT, Netflix for TTML, and American TV for SCC. If a broadcaster asks you for captions, they will name the format and usually the vendor.

SRT vs VTT: what is actually different?

Short answer: very little that you can see. Both are plain-text files of numbered or unnumbered cues, each with a start time, an end time and the words. VTT adds a header line, a full stop instead of a comma in the times, and optional extras such as screen position and speaker tags. But VTT is not "a bigger SRT": an SRT file is not a valid VTT file, and a renamed one will fail.

Here is the same caption in both formats. First, SRT:

1
00:01:05,200 --> 00:01:08,900
The meeting point moves
to the lawn by the main gate.

And VTT:

WEBVTT

00:01:05.200 --> 00:01:08.900 line:10%
<v Maria>The meeting point moves
to the lawn by the main gate.

The differences, one per line:

• The first line. A VTT file must start with WEBVTT; an SRT file starts with the first cue's number (W3C).
• The decimal mark. SRT writes 00:01:05,200 with a comma; VTT writes 00:01:05.200 with a full stop.
• Hours. SRT always has them; VTT may leave them out under an hour (01:05.200).
• Numbers. SRT numbers every cue from 1; in VTT a cue label is optional.
• Position. VTT can place a caption, for example at the top of the screen with line:10%, so it does not cover a speaker's name on a slide. SRT has no standard way to do that.
• Speakers. VTT can tag who speaks, <v Maria>, which Microsoft Teams uses in its transcripts. In SRT a name is just text.
• Notes and styling. VTT allows NOTE comments and a STYLE block; SRT's bold and italic tags are unofficial, and YouTube ignores them (YouTube).

Where they came from. SRT is named after SubRip, a free Windows program from 1999 that read subtitles off the picture of a DVD and saved them, with their timings, to a text file (Wikipedia). Nobody ever published a standard for it. The US Library of Congress notes "little to no information on SRT formal documentation" and no official character encoding (Library of Congress). That is why one SRT plays everywhere and the next one is refused.

VTT began in 2010 as "WebSRT", a draft for captions in HTML5 video. Silvia Pfeiffer, who later edited the standard, described it then as "an extension to the existing SRT SubRip format" (Pfeiffer, 2010). In January 2011 a change to the draft titled "Rename SRT to VTT." gave it its present name (W3C archive). Sixteen years on, WebVTT is still a W3C Candidate Recommendation Draft, dated 20 May 2026, not a finished standard (W3C). It works in every modern browser all the same.

SRT or VTT: which one should I use?

Short answer: VTT for a web page, Moodle, Rise 360, and Canvas viewers on iPhones; SRT for Facebook, LinkedIn, X, Premiere Pro and Final Cut Pro; either for YouTube, Vimeo, Storyline 360, Captivate, Kaltura and Panopto. But do not choose blind if you post to both kinds: keep both files. Converting one into the other takes seconds and loses almost nothing.

When only one file will do, SRT is accepted by more upload forms, and VTT by more web players and course tools. A teacher who puts the same lecture on YouTube and in Moodle needs both.

What converting loses. SRT to VTT keeps every time and every word; only the unofficial font-colour tags, which few files have, are dropped. VTT to SRT drops the screen positions, NOTE comments and STYLE blocks, and the speaker tags become plain text. If your VTT places captions at the top of the screen on purpose, the SRT copy will put them back at the bottom.

By hand, SRT to VTT is four steps:

1. Add WEBVTT as the first line, then an empty line.
2. In the timing lines only, change the comma before the milliseconds to a full stop. Do not search-and-replace every comma: the captions have commas too.
3. Leave the numbers; VTT reads them as labels.
4. Save as UTF-8 with the .vtt ending.

The other way round, delete the header, NOTE and STYLE blocks and anything after the end time; add 00: where the hours are missing; change the full stops back to commas; number the cues from 1. The converter below does both directions and lists what it changed.

VTT vs DOCX or TXT: do I need captions or a transcript?

Short answer: for a video with sound, captions: a timed SRT or VTT. For an audio-only recording, such as a podcast or a recorded talk with no picture, a transcript: DOCX, PDF or TXT. Give both if you can: they serve different people. But a transcript of a video does not replace its captions, however good it is.

The international accessibility guidelines, WCAG, say it plainly at their basic level, A:

But the laws below ask for level AA, which goes one step further: at AA, a video whose picture matters needs audio description, and a transcript alone is no longer enough (WCAG 1.2.5). A lecture where the speaker says "as you can see on this slide" is that kind of video.

Captions, in WCAG's definition, are a synchronized alternative: the words appear while they are spoken, and they also name the speaker and the sounds that matter. A Word file cannot do that.

The case teachers should know. In February 2015 the US National Association of the Deaf sued Harvard and MIT over online videos with no captions or with inaccurate ones (Cohen Milstein). Harvard settled in November 2019; MIT in February 2020, agreeing to caption the content it posts. The association's chief executive, Howard Rosenblum, asked the question that sums up the case: "Why would you not make your content accessible to everyone?" (AP, 2020).

The rules where you teach, as of October 2026:

• United States, public schools, colleges and universities: the ADA Title II rule of 2024 requires WCAG 2.1 AA, captions included. On 20 April 2026 the Department of Justice moved the deadlines back a year: 26 April 2027 for bodies serving 50,000 people or more, 26 April 2028 for smaller ones (ADA.gov; UNC School of Government). The department also said it plans further rulemaking on the requirements themselves, so check before you plan around them (Federal Register).
• United Kingdom, public sector: WCAG 2.2 AA; video published before 23 September 2020 is exempt, and schools and nurseries are partly exempt (GOV.UK).
• European Union, public bodies, public universities among them: the Web Accessibility Directive, with the same 23 September 2020 cut-off for video as the UK; countries may exempt schools and nurseries (Directive 2016/2102). The European Accessibility Act, which has applied since 28 June 2025, covers services sold to consumers, such as e-books and online shops; it does not name education (Directive 2019/882). Ask your institution which one covers your course.

How do I turn a Zoom or Teams VTT into a Word document?

Short answer: in Teams, skip the VTT: the meeting organiser can download the transcript as a .docx. In Google Meet, the transcript is already a Google Doc. In Zoom you get a VTT, and opening it in Word leaves every timestamp and every broken line in place: strip them, join the lines into paragraphs by speaker, and keep the names. But read it before you share it: a recorded class carries your students' names.

What each meeting tool gives you, as of October 2026:

• Zoom (cloud recording, paid plans): "Audio transcripts are saved in VTT format" (Zoom).
• Microsoft Teams: organisers and co-organisers "can download it as a .docx or .vtt file"; the VTT tags each speaker with <v Name> (Microsoft; the tags).
• Google Meet: a Google Doc in the organiser's Drive, on Workspace plans that include transcripts; no caption file (Google; Google, 2022).

A Zoom VTT holds one fragment of a sentence every few seconds, each above its own timing line. One developer on Zoom's forum summed up working with it: "This is inconvenient to parse, modify- extract etc." (Zoom Developer Forum, 2022). Colleges that answer the question tell staff to open the file in Word, timestamps and all (CCAC), or to rename it to .txt and ask an AI assistant to clean it up (Hebrew Union College). The second route works, but it sends a class recording's names to another service.

The converter below does the cleaning in your browser. Paste or choose your file, pick A readable transcript, and download it as Word or text. It joins each speaker's lines into paragraphs, keeps the names, and can keep a timestamp every 2 or 5 minutes so you can still find a moment in the recording. Choose A caption file to convert SRT to VTT or back, fix broken times and check the lines.

The .txt it makes is also the file to give an AI tool: no timestamps to confuse it, and the speaker names still there.

How do I turn text in Word into an SRT?

Short answer: you cannot by renaming it. A caption file needs a start and an end time for every line, and a Word document has none. The timings have to come from the recording: let YouTube align your text to the audio, run the recording through a transcription tool, or time it yourself in a free editor. But each route has a catch, listed below.

• YouTube Auto-sync. Save the text as plain .txt (in Word, Save As, then Plain Text) and upload it to your video as a transcript; YouTube times it to the speech (YouTube). The catch: the text must be in the video's language, and one YouTube's speech recognition supports; YouTube does not recommend it for videos over an hour or with poor sound.
• A transcription tool. Upload the recording, check the text, export SRT. In Kukarella, open TranscribeHub, upload the file and press Transcribe; when the text appears, Export → Srt → Download. Transcription uses one minute from your balance per minute of recording; exporting is free. The catch: our SRT is built from the timings of the first transcription, so a word you correct in the editor stays uncorrected in the SRT. Correct it in the caption file, or use our Txt or Doc for the corrected text.
• Subtitle Edit, a free, open-source subtitle editor for Windows, with builds for macOS and Linux, lets you time lines against the sound wave and can also transcribe with Whisper, OpenAI's speech-recognition model (Subtitle Edit; GitHub). The catch: it takes an evening to learn.

If you already have a script and a finished video, YouTube is the quickest. If you have only the recording, a transcription tool is.

Why won't my caption file upload?

Short answer: almost always a small format slip, not the words. Check the encoding, the first line, the decimal marks and the blank lines, in that order. But the fault is often invisible in a text editor, which is why the converter's check exists.

A thread on Adobe's forum, "Captions (.srt) will not import", has collected replies since 2022. An Adobe community expert, Stan Jones, put the usual cause in one line: "almost always an extra line or some other formatting" (Adobe Community). One user in the thread fixed it by changing the line endings and saving the file as UTF-8.

The checklist:

1. Encoding. Save as UTF-8. YouTube asks for "plain UTF-8", Vimeo requires it, and Articulate says that letters showing up as symbols mean the file is not UTF-8 (Articulate).
2. The first line. A VTT without WEBVTT at the top is refused; Rise 360 says so in its help.
3. Decimal marks. Commas in SRT times, full stops in VTT times.
4. Blank lines. One empty line between cues, none inside a cue.
5. The file name, for Facebook: name.en_US.srt, language in small letters, country in capitals (Facebook); Meta's help for video ads says captions will not upload otherwise (Meta).
6. Times past an hour. Some tools write the minutes wrong after the first hour, 01:61:40 instead of 01:01:40. Ours is one of them, sometimes: TranscribeHub's SRT export makes this mistake when Auto break is switched off in Reflow Subtitles, or when our subtitle server cannot be reached or turns the request down, and it does not tell you when that happens. The fix is filed, and until it ships the converter above reads such times correctly and rewrites them.

How long should a caption line be, and how fast?

Short answer: two lines at most, about 42 characters a line, on screen long enough to read. For classroom video, the US Described and Captioned Media Program, funded by the Department of Education, sets the pace by level: no more than 130 words a minute for lower levels, 140 for middle and 160 for upper (DCMP). But most auto-captions ignore these limits, so check before you publish to a young class.

Sources: Netflix English (USA) style guide, BBC subtitle guidelines, DCMP Captioning Key. DCMP also keeps each caption on screen for at least 40 frames, about 1.3 seconds, and no longer than 6 seconds.

The converter's caption mode flags lines over 42 characters and cues faster than 20 characters a second, or 17 for children, and breaks long lines for you. But it does not change the timings: a cue that runs too fast must be split or lengthened by hand.

Getting transcripts and captions out of Kukarella

Short answer: TranscribeHub exports four files and charges nothing for exporting, even without an account. But none of the four is VTT, the text files carry no timestamps or speaker names, and the SRT has the faults described above.

In a TranscribeHub project, Export opens Select File Format with four choices:

• Doc: a real Word .docx of the transcript text as you edited it. No timestamps, no speaker names, no title.
• Pdf: the same text as a PDF.
• Txt: the same text as plain UTF-8, one line per paragraph. But a fresh transcript is one paragraph per recording until you break it up in the editor.
• Srt: captions built from the transcription's timings, not from your edits.

TranscribeHub exports four files: Doc, Pdf, Txt and Srt.

With Srt chosen, Configure... opens Reflow Subtitles, where you set how many characters and lines a caption holds and how fast it may run, or pick a preset: Broadcast, YouTube or Social Shorts. One catch: if you export without opening this panel, the SRT allows up to 84 characters a line, twice Netflix's limit. For a class, open it and set Max characters per line to 42.

The Broadcast preset: up to 84 characters, but no more than 5 words, a line.

What it does not do, so you can plan around it:

• No VTT. Convert the SRT with the tool above.
• No speaker names. Our transcription does not tell speakers apart; add names by hand.
• Edits do not reach the SRT. Correct the caption file itself.
• SRT past one hour: see point 6 of the checklist.
• Text-to-Voice and DialoguesAI export audio (MP3 or WAV) and the script as TXT or DOC, but no caption file.

Quick check: which file?

Three situations teachers and trainers meet. Choose an answer to see why.

Questions people ask

What does SRT stand for?

SubRip Subtitle, after the SubRip program. It is not "Secure Reliable Transport", which is a video-streaming protocol with the same letters; even X's caption help mixes the two up.

What is a VTT file, and how do I open it?

A plain-text caption file, Web Video Text Tracks. It opens in Notepad, TextEdit or Word (choose All Files in the open window), but you will see every timestamp. To read it as a transcript, convert it first.

Is VTT better than SRT?

For a web page and some course tools, yes, because they take nothing else. For Facebook, LinkedIn, X or a video editor, no, because they take only SRT. For YouTube and most course tools, either works.

Can I rename a .txt or .docx file to .srt?

No. The ending changes, the content does not: without numbered cues and timing lines, no player can read it. Renaming .srt to .vtt fails too, because of the missing header and the commas.

Can ChatGPT read a VTT file?

OpenAI's help lists DOCX, PDF and TXT for documents and text; VTT is not on its list (OpenAI). Turn the VTT into plain text first, which the converter above does, and keep the speaker names.

Which format does YouTube want?

Either SRT or VTT, in UTF-8. YouTube also takes a plain-text transcript and times it itself.

DOCX or PDF for a transcript?

DOCX if anyone will edit, comment on or reuse it; PDF for a fixed record to print or file. Neither is captions.

Does Kukarella export VTT?

No. TranscribeHub exports Doc, Pdf, Txt and Srt; the converter in this article turns the SRT into a VTT in your browser.

How we checked

Every format rule comes from its specification or the publisher's help page, and every platform list from that platform's own help, read on 18 October 2026. Kukarella's exports were read in our product's code that day and checked in the product on 28 October 2026. We did not test uploads on every platform listed; where a help page and real uploads disagree, trust the upload and tell us.

Sources

All pages read on 18 October 2026 unless stated. Platforms change their lists often.

• W3C. WebVTT: The Web Video Text Tracks Format, Candidate Recommendation Draft, 20 May 2026. w3.org/TR/webvtt1
• W3C. Media Accessibility User Requirements, Working Group Note, 3 December 2015. w3.org/TR/media-accessibility-reqs
• W3C. Understanding WCAG 2.2: success criteria 1.2.1, 1.2.2, 1.2.3, 1.2.5.
• W3C public-html-diffs archive, "Rename SRT to VTT.", January 2011. lists.w3.org
• Silvia Pfeiffer, WebSRT and HTML5 media accessibility, 7 August 2010. gingertech.net
• Library of Congress, SubRip Subtitle format (SRT). loc.gov
• Wikipedia, SubRip. wikipedia.org
• MDN, The track element. developer.mozilla.org
• Platform help pages: YouTube formats, YouTube Auto-sync, YouTube transcripts, Google Drive, Vimeo, Facebook and its naming rule, LinkedIn, X, Canvas, Moodle, Rise 360, Storyline 360, Captivate, Kaltura, Panopto, Premiere Pro, Final Cut Pro, Zoom, Microsoft Teams, Microsoft Graph, Google Meet.
• Caption guidelines: Netflix English (USA) Timed Text Style Guide; Netflix General Requirements; BBC Subtitle Guidelines; DCMP Captioning Key, text and presentation rate.
• Law: ADA.gov, the Title II web rule; Federal Register, 20 April 2026; Directive (EU) 2016/2102; Directive (EU) 2019/882; Kristi Nickodem, UNC School of Government, 20 April 2026; GOV.UK, public sector accessibility; European Commission, European Accessibility Act.
• The Harvard and MIT cases: Cohen Milstein case page; AP, 18 February 2020.
• Forums and help centres: Adobe Community, "Captions (.srt) will not import"; Zoom Developer Forum, 2022; CCAC Help Center; Hebrew Union College.
• Subtitle Edit. nikse.dk/subtitleedit; GitHub
• OpenAI, Uploading files and audio to ChatGPT. help.openai.com