Transcribe a Zoom, Google Meet, or Teams recording

Step one: find where your platform actually saved the recording

Each meeting platform stores recordings in a different place, and the location depends on whether you recorded locally or to the cloud. Zoom local recordings are saved to your own machine, by default in a Zoom folder inside your Documents directory, and typically include both an MP4 video file and a smaller M4A audio-only file. Zoom cloud recordings live in your account's web portal under the Recordings section rather than on your hard drive.

Google Meet recordings are saved to Google Drive, specifically to the meeting organizer's Drive, usually in a folder named Meet Recordings, as MP4 files. If you attended the meeting but did not organize it, you may only have a shared link rather than the file itself, which matters for the next step. Recording availability also depends on the organizer's plan, so not every Meet call will have produced a file at all.

Microsoft Teams stores meeting recordings in OneDrive for ordinary scheduled meetings (in the recorder's account) or in SharePoint for channel meetings, again as MP4 files. Teams recordings in many organizations are set to expire automatically after a retention window set by an administrator, so it is worth retrieving the file sooner rather than later, a transcript is only possible while the recording still exists.

Download a real copy of the file before doing anything else

Whatever transcription route you choose, you want the actual media file, the MP4 or M4A, saved locally on your computer, not just a streaming link in a browser tab. Transcription tools work on uploaded files, and a share link that plays a recording in a web player is not the same thing as the file itself. On each platform there is a download option: in Zoom's web portal next to each cloud recording, in Google Drive via the download command on the file, and in OneDrive or SharePoint through the file menu in Teams or the web interface.

Permissions can get in the way here. Frequently only the meeting organizer or the person who started the recording has download rights, and some organizations disable downloads for shared recordings entirely. If you find yourself able to play a recording but not save it, the honest fix is to ask the organizer to download the file and send it to you, or to grant you download access, not to hunt for workarounds that may violate your workplace's policies.

Having a local copy also protects you against expiry. Cloud recording storage is limited or time-boxed on several platforms, and more than a few people have gone looking for an important recording weeks later only to find it gone. Once the file is on your machine, back it up like any other document that matters.

Want to try it now? Upload a file to FastScribe. Your first one is free, no signup.

Pick your route: manual typing, built-in transcripts, or an automated service

The do-it-yourself route costs nothing but time, and a lot of it. Transcribing by ear typically takes several times the length of the recording even for a fast typist, because you will pause, rewind, and replay constantly. It makes sense for very short clips, for material too sensitive to leave your machine, or when you need to deeply absorb the content anyway. For a full hour-long meeting, most people find it impractical.

Some meeting platforms and workplace plans generate their own transcripts automatically, either during the call or attached to the cloud recording afterward. If yours does, check there first, it may save you the whole exercise. The catch is that this feature is often gated behind specific subscription tiers or administrator settings, the output format may be awkward to reuse, and quality varies. If the built-in option was not enabled before the meeting happened, it usually cannot be applied retroactively to a saved file.

Automated transcription services are the middle path: you upload your recording, software converts the speech to text, and you get a downloadable transcript back, generally in a small fraction of the recording's runtime. Quality on clear meeting audio is good enough that cleanup is a light editing pass rather than a rewrite. This is the route most people take for routine meetings.

Professional human transcription services occupy the high end. A person listens and types, which handles heavy accents, crosstalk, poor audio, and specialized terminology better than software does. The trade-offs are turnaround measured in days rather than minutes and a meaningfully higher cost. For board minutes, research interviews, or anything where every word must be right, they earn their keep; for a weekly status call, they are usually overkill.

Prepare the file so the transcript comes out cleaner

A little preparation improves both speed and results. First, trim the junk: many meeting recordings open with minutes of waiting-room silence, screen-share fumbling, or small talk before the meeting proper begins. Cutting that dead air with any basic video or audio editor means less to upload, less to process, and less noise in your transcript.

Second, consider extracting just the audio track. The words in an MP4 meeting video live entirely in its audio, and an audio-only M4A or MP3 file is often a tenth the size of the video or smaller. A free command-line tool like ffmpeg can copy the audio out of a video without re-encoding it in a few seconds, and many desktop editors can do the same. A smaller file uploads faster and fits within tighter size limits, and the resulting transcript is identical.

Third, set expectations based on the audio you have. Speech-to-text of any kind, human or machine, degrades when people talk over each other, when someone's microphone was far away or muffled, or when the recording captured a laptop speaker rather than direct audio. You cannot fix that after the fact, but knowing it helps you budget proofreading time, and it is a good argument for asking colleagues to use headsets in meetings you plan to transcribe.

Choose the right output format for what you will do with the text

Plain text (TXT) is the workhorse. It opens anywhere, pastes cleanly into email, documents, and wikis, and is the right choice when the goal is meeting minutes, action items, or a searchable record. If you only ever use one format, this is it.

SRT and VTT are timestamped caption formats. Instead of one continuous body of text, they break the transcript into short timed segments that a video player can display in sync with playback. Choose one of these when you plan to share the recording itself and want it subtitled. SRT enjoys the widest support across desktop players and video platforms, while VTT is the standard for web-based players. Because they carry timestamps, they are also handy for jumping to the exact moment something was said.

Word-processor formats such as DOCX matter when the transcript is headed into an editing or review workflow, when someone will restructure it, add comments, or track changes before it circulates. Many services let you export the same transcript in multiple formats, so you rarely have to commit up front; think instead about the destination for each copy.

Proofread and clean up the machine output

No automated transcript should be treated as finished. Software reliably stumbles on the same categories of words: people's names, company and product names, acronyms, and domain jargon. If your meeting discussed a project with an invented codename, expect it to be spelled creatively. A focused pass with the recording playing at increased speed lets you correct these quickly, most people can review far faster than they could ever have typed the transcript from scratch.

Beyond fixing misheard words, cleanup is where the transcript becomes genuinely usable. Break the wall of text into paragraphs at natural topic shifts, add speaker names where knowing who said what matters, and decide whether to strip filler words and false starts. A verbatim record and a readable record are different documents; know which one you need before you start editing.

Do the cleanup soon after the meeting while your memory can resolve ambiguities. Two weeks later, you will not remember whether the garbled phrase at minute 34 was a budget figure or a date, the day after, you often will.

Think about privacy before you upload

A meeting recording contains your colleagues' voices and, quite possibly, confidential business information. Before uploading it anywhere, it is worth a moment's thought about whether that is appropriate for this particular recording. Most workplaces have policies about where company data may be sent; if yours does, follow it, and if you are unsure whether a specific recording is sensitive, ask the meeting owner rather than guessing.

When evaluating any transcription service, read what it actually does with your upload. The questions that matter: how long is the audio kept, is your content used to train models, can you delete it, and where is the processing performed. Services differ widely here, and the answers are usually in the privacy policy or a data-handling page, absence of an answer is itself an answer.

For genuinely sensitive material, remember that fully offline options exist: open-source speech-recognition software can run on your own computer so the recording never leaves it. Setup takes more effort and a reasonably capable machine, but for some content that trade is the right one.

Where FastScribe fits

Where does FastScribe sit among these options? It is one of the automated services described above, not a human transcription service, and not a substitute for typing it yourself when a recording is too sensitive to upload. Concretely: you upload your meeting file (there is no fetching from a link or connecting an account, so download the recording first as described earlier), it is transcribed with our own transcription engine running on our own servers, and you can export TXT, SRT, or VTT. Your first file is free with no signup for recordings up to 50 MB and 10 minutes, enough to test quality on a clip of your meeting, though most full meetings will exceed it. A free account raises the file-size limit to 100 MB. The Pro plan at $12 per month accepts files up to 2 GB and 5 hours, adds DOCX export, batch uploads, a priority queue, and permanent history, with unlimited files under fair use. On data handling: uploaded audio is deleted immediately after transcription; the transcript text is kept so you can retrieve it later. If your platform already generated a good transcript, use that; if the audio is rough or the stakes are high, a human service may serve you better. For the common case, a clear recording you need in writing today, an automated tool like FastScribe is the practical middle ground.