Microsoft Teams stops accepting a caption file you upload yourself once it passes 4MB. The number does not live in a specifications table. It appears in one place only — the list of errors on Microsoft's page about managing video transcripts and captions: "Your file size exceeds the 4MB limit that this feature supports."
That is why almost nobody knows the cap exists until they hit it, usually after they have already corrected, translated or converted a subtitle file and are trying to attach it to a video.
This page covers what the 4MB actually measures, the other caption-related numbers Teams documents (they are different numbers for different things, and mixing them up is the real source of confusion), and what to do when your file is over the line.
The limits, at a glance
| What is being measured | Documented value | Where Microsoft states it |
|---|---|---|
| A captions/transcript file you upload to a video | 4MB | error list: "Your file size exceeds the 4MB limit that this feature supports." |
| Format accepted on that upload | WebVTT | "Upload your own WebVTT captions and transcript file by selecting Video settings on the right-hand side of a video's player page." |
| Video size and length for automatic transcript generation | "1GB and 4 hours" in one section — and "less than 4GB in size and less than 4 hours" in another section of the same page | "Size limit and possible failures" and "Supported file size" |
| Meeting recording maximum length | "4 hours or 1.5 GB. When this limit is reached, the recording ends and automatically restarts." | limits and specifications page |
| Official estimate of recording size | "The size of a one-hour recording is 400 MB." | recording storage planning page |
Two things to read carefully in that table.
The 4MB is about the subtitle file, not the video. It belongs to the manual upload path — the one where you open a video's player page, expand the transcripts and captions section, and supply your own file.
The 1GB and the 4GB both appear on the same support page. One section gives the limit for running automatic transcripts as "1GB and 4 hours"; a later section on the same page says generation "are only supported for videos less than 4GB in size and less than 4 hours." The duration agrees at 4 hours in both places. Plan against the smaller of the two size figures, because the cost of guessing wrong is a generation job that simply fails.
Why the limit exists where it does
Teams does not treat a caption file as a normal document. Microsoft describes it as something bound to the video: "The captions and transcript file is stored with the video in OneDrive for Business or SharePoint. This inherits the same permissions and policy as the video file, so that whoever has access to the video file has access to the captions and transcript."
It is also deliberately not exposed as a file you can manage: "There is no direct access to see or manage the captions and transcript file in the file explorer for OneDrive for Business or SharePoint. Because this file only enhances the video viewing experience, it is viewed and managed only through the video player as described above."
So the 4MB is not a storage quota — your OneDrive has room for a file this size many times over. It is a ceiling on an attachment the player is expected to parse and render in real time.
What actually pushes a caption file toward 4MB
A WebVTT file is plain text, and Microsoft's own description of the format explains where the volume comes from: "The first line must read WEBVTT and subsequent entries in the file are called 'cues'." Every cue carries a time marker and its text. More speech, more cues.
Three documented details decide how many cues you end up with:
- Duration. Microsoft's accepted time marker formats are
hh:mm:ss.ttt,hhhh:mm:ss.ttt, ormm:ss.ttt— note that a four-digit hours field is allowed at all. For context on how long a single session can legitimately run, Microsoft states that "Microsoft Teams meetings and events (such as webinars and town halls) have a time limit of 30 hours." - Cue density. The guidance for the text of each cue is "Limit text to a maximum of three lines to ensure all the words will fit on the screen." Shorter cues on screen means more cues in the file for the same amount of speech.
- Extra fields do not help you. Microsoft is explicit that "Additional fields, such as notes, comments, styles, etc., are not supported and will be removed from the file upon upload." The documentation describes the removal, not the arithmetic — our recommendation is to treat the size of the file as it sits on your disk as the number that counts, rather than hoping stripped styling brings you under the cap.
What to do when you hit it
If the file came from Teams, it is already WebVTT. Microsoft's live transcript page states that "By default, meeting organizers and co-organizers can download it as a .docx or .vtt file." There is no SRT export — the details of who is allowed to download which one are in our guide to downloading a Teams transcript as .docx or .vtt.
If the file is an SRT, converting is not optional. The upload dialog is documented as taking a WebVTT file, and renaming an .srt does not change the timecodes inside it. Microsoft spells out the timing syntax it expects: "After ss, include a period (.) and three digits to represent thousandths of a second (ttt)." A file whose timecodes do not match that will still be wrong after a rename.
Try it right here
Runs in your browser. Your file is not uploaded anywhere.
This runs inside your browser tab. The file is read from disk, rewritten locally, and handed straight back — nothing is uploaded to a server, which matters more than usual for a meeting transcript, since it is a verbatim record of what named people said. The same SRT to VTT converter is what you want any time a player refuses your subtitles because it only speaks WebVTT.
If it is still over 4MB after conversion, cut scope. Uploading captions for the section people actually need is the only lever entirely under your control. Microsoft's page does not describe a way to raise the cap, so check the official documentation before assuming an admin setting exists for it.
Platform-specific gotchas
- The file is invisible in OneDrive. Do not go looking for a subtitle file beside the MP4 — per the quote above, there is no direct access to it in the file explorer, only through the video player.
- Copying the video can break the captions, and two official pages scope that differently. The admin page says the transcription link "can be broken if the video file is copied within the same OneDrive or SharePoint site. This action would result in captions not displaying on the copied video file." The storage page puts it more broadly: "Closed captions aren't fully supported if the Teams meeting recording is moved or copied from its original location in OneDrive or SharePoint."
- Live captions are never a source file. Microsoft's answer to whether Teams stores caption data is "No. Caption data is permanently deleted after the meeting is over." Whatever you upload has to come from a transcript or from your own tooling.
- Editing costs you speaker names. "Teams live transcripts edited in this way lose the speaker attribution of who said what during the meeting."
FAQ
What is the caption file size limit in Microsoft Teams? 4MB for a captions or transcript file you upload yourself. Microsoft documents it as the error "Your file size exceeds the 4MB limit that this feature supports." It applies to the upload path on a video's player page, not to the video itself.
Can I upload an SRT file as captions in Teams? The documented upload is for a WebVTT file. Convert first — and convert properly, since the WebVTT timing format Microsoft describes uses a period and three digits for thousandths of a second.
Is the 4MB limit the same as the 1GB transcript limit? No. They measure different things. The 4MB is the subtitle file you supply; the 1GB (and, in another section of the same page, 4GB) applies to the video Teams runs automatic transcription on. More Teams numbers are collected in our limits hub.
Why does my long meeting have no automatic captions? Automatic generation is documented as limited to 4 hours, and by size. With Microsoft's own estimate of 400 MB per recorded hour, a long session runs into the recording split at "4 hours or 1.5 GB" as well — see the 4-hour and 1.5 GB recording limit.
Will Teams shrink my caption file for me? Not in a way you can plan around. Microsoft documents that unsupported fields "will be removed from the file upon upload," but says nothing about that happening before the size check. Get the file under 4MB yourself.
Quotes from Microsoft's official documentation on viewing, editing and managing video transcripts and captions, live transcripts in Teams meetings, live captions, recording storage and permissions, and the Teams limits and specifications page, checked 17 September 2026. Microsoft revises these pages — verify before relying on a number. Related: SRT to VTT, all Clapr tools.
