Guide

How to transcribe a Zoom, Teams or Google Meet recording

Your meeting platform may already offer a transcript — on the right plan, in the right languages, stored on their servers. Here is when to use it and when not to.

30 August 2026

Every major meeting platform now offers transcription, and every one of them attaches conditions: a paid tier, a supported language list, cloud recording rather than local, and a copy of the conversation kept on their infrastructure.

Often that is fine. Sometimes it is not, and then you need the recording file itself.

Where your recording actually is #

Before anything else, find the file. This is the step people get stuck on.

Zoom. Local recordings are written to a folder on your machine when the meeting ends — on a Mac, usually ~/Documents/Zoom/, in a subfolder named for the meeting. You will typically find an MP4 of the video and a separate M4A of the audio. Either works for transcription; the M4A is smaller. Cloud recordings live in your Zoom account instead, and can be downloaded from the web portal.

Microsoft Teams. Recordings usually go to OneDrive (for a regular meeting) or SharePoint (for a channel meeting), as an MP4. You can download the file from there.

Google Meet. Recordings are saved to the organiser's Google Drive, in a "Meet Recordings" folder, as an MP4.

Once you have the file, everything below applies equally to all three.

When the built-in transcript is the right answer #

Use it when all of these are true:

• You are on a plan that includes it.
• The meeting is in a language the platform supports well.
• The recording is already in their cloud.
• Nobody in the meeting would object to the transcript living there.

For an internal standup, that is a perfectly good deal and you should take it.

When to transcribe the file yourself #

The meeting was recorded locally. A local Zoom recording never reaches Zoom's servers, so there is nothing for their transcription to work on unless you upload it — which defeats the reason you recorded locally.

The plan does not include it. Transcription is a paid feature on all three platforms. If you are on a free tier, the file on your disk is your route.

The language is not well supported. Platform transcription covers a much shorter list of languages than Whisper's 99, and quality on the less common ones varies. This is the most common reason people give up on built-in transcription.

The conversation is confidential. Client work under NDA, a candidate interview, a legal or medical discussion, an investigation. The question is not whether the platform is trustworthy — it is whether you can accurately describe where the recording went, if someone asks. "It is on my laptop" is a complete answer. "It is transcribed by our conferencing vendor and retained under their policy" is a longer one.

You need subtitles. Built-in transcripts are usually a text panel or a plain document. If you need an SRT or VTT to caption a recording you are publishing, you want a tool that exports subtitle files.

The recording is old. Retention policies delete cloud recordings. A downloaded MP4 from two years ago has no platform transcript available and never will.

Doing it locally #

The mechanics are simple, because the MP4 already contains everything needed:

1. Open Offline Transcription and drop in the MP4 or M4A. The audio track is extracted automatically — you do not need to convert the video first.
2. Set the language if it is not English. For a multilingual meeting, pick the dominant language; auto-detection samples the start of the file and can settle on the wrong one if the first minute is small talk in another language.
3. Transcribe. A one-hour meeting takes a couple of minutes on Apple Silicon.
4. Export as TXT for notes, SRT or VTT if you are captioning the video, or CSV if you want to sort and search segments in a spreadsheet.

Nothing is uploaded, which also means it works on a plane or in an office with no outside network access.

Speaker names, honestly #

Built-in transcription has one genuine advantage worth naming: the platform knows who was speaking, because it knows who was unmuted. That produces speaker labels no post-hoc tool can match.

Automatic speaker separation from a mixed audio file is a different and much harder problem, and every tool that offers it — including the ones that charge for it — is unreliable when several people talk over each other. If accurate speaker attribution matters more than anything else, and the meeting was cloud-recorded on a paid plan, use the platform's transcript.

A note on recording meetings at all #

Tell people. Every platform announces it, which is the point.

Beyond the platform notice, recording a conversation has legal implications that vary considerably by country and by state, particularly for anything resembling a phone call. If you are recording something sensitive, check what applies where you and the other participants are — that is genuinely worth thirty seconds before, rather than an argument after.

The short version #

Find the file first — Zoom writes local recordings to ~/Documents/Zoom/, Teams to OneDrive, Meet to Drive. Use the built-in transcript when you are on a plan that has it and the conversation is not sensitive. Transcribe the file yourself when the recording is local, the language is unusual, the content is confidential, or you need actual subtitle files.