How to transcribe a Zoom recording with speaker labels

By Alex Topilski, Founder

2026-07-25

Zoom's own transcription lives behind two paywalls at once: cloud recording and its transcript feature require a paid plan — Pro is $14.16 per user per month billed annually ($16.99 monthly) — and even then the transcript sits in Zoom's cloud, not in a file you control. What free Zoom does give every host is local recording: a real video and audio file saved to your own computer. That file is all you need. This guide takes it from the Zoom folder to a transcript with named speakers, free, in about the time it takes to read the meeting's action items.

1. Record the meeting locally

In the Zoom desktop app, press Record in the toolbar — or Alt+R on Windows, Cmd+Shift+R on Mac. On the free plan this is local recording (cloud recording is the paid feature), and it captures every participant's audio, not just your side. Two constraints to know before the meeting starts:

• Recording is host-gated. The host can record; a participant only gets the Record button if the host grants permission (Participants panel → More → Allow Record). Ask before the meeting, not during it.
• Desktop only. Local recording works in the Windows, macOS and Linux clients, not in the mobile apps. Free-plan group meetings also end at 40 minutes — for a longer session, restart the meeting and you simply get a second file to transcribe.

When the meeting ends, Zoom converts the recording — a progress window appears — and drops the files into Documents/Zoom, in a subfolder named after the date and meeting topic. Inside you will find a video file (.mp4) and an audio-only file (audio_only.m4a). Both contain the same conversation; the .m4a is a fraction of the size, so it uploads faster. An hour-long meeting is typically 20–60 MB as audio against hundreds of MB as video.

If you were a participant who could not get record permission, there is a browser route: join the call with Zoom's "join from your browser" option instead of the desktop app, and the free Meet Voice Recorder extension records the call — every remote participant plus your own microphone — straight to a .webm in your Downloads folder, which uploads here as-is. It works only for browser-joined calls, not the desktop app. And ask first anyway: the consent conversation (below) is required no matter who presses Record.

Get a cleaner recording before it starts

Transcript quality is mostly decided during the meeting, not after it. Three habits pay for themselves:

• Headsets beat laptop microphones. A built-in mic picks up keyboard clatter, fan noise and room echo along with the voice; a headset or USB microphone captures the voice alone. One participant on a bad mic degrades only their own lines — but if that participant is the one making the decisions, those are the lines you needed.
• Mute when not speaking, and take turns. Overlapping speech is the hardest case for both word accuracy and speaker attribution: even 2 seconds of two voices at once can push the next line to the wrong speaker. Zoom's mute discipline, which meetings need anyway, is exactly the discipline transcription rewards.
• Start the recording before the first agenda item. Decisions have a habit of being made in the first two minutes, while someone is still saying they will "start recording in a second".

2. Upload the file

Drop audio_only.m4a — or the .mp4, if that is what you have — on the WhoSayAh home page. Both are accepted as-is; video containers work because only the audio track is read. There is no account and no queue to join: files up to 100 MB and 60 minutes are transcribed free, and both limits are checked at upload time, so an over-limit file is rejected in seconds, before any waiting.

The upload dialog shows the file's name, size and detected length, and asks two things:

Language. Leave Detect automatically on unless the meeting was short or noisy — with less clean audio to work from, naming the language removes one source of error. For a bilingual meeting, pick the language that dominated.

Speaker count. Keep Label speakers on and enter how many people actually spoke — not how many attended. A 10-person all-hands where 3 people talk is a 3-speaker recording. If you genuinely do not know, I don't know — detect it works, but the exact count is the single setting that most improves attribution. Then press Transcribe.

3. Name the speakers, download the transcript

Lines appear as they are transcribed — a 30-minute meeting is typically readable within 3 to 5 minutes — each prefixed Speaker 1, Speaker 2. When the transcript completes, press Name speakers: each voice is listed with how long it spoke and its first line, which for a meeting you just attended is enough to place everyone. The names apply across the whole transcript at once.

Then take the text in the form the job needs:

• Copy — one click, paste into minutes, a ticket, or a follow-up email.
• .txt — one line per speaking turn, name prefixed; the file to attach to the project record.
• .vtt — timestamped WebVTT subtitles. If you keep the .mp4, this file makes it a captioned video: the subtitles guide shows where the file plugs in, from YouTube to your own site.

Why not just use Zoom's built-in transcript?

If your organisation already pays for Zoom and the meeting was cloud-recorded, use the built-in transcript — it is included in what you are paying for. The upload route earns its place in the cases that paywall leaves out:

• You are on the free plan. Local recording works; cloud transcription does not exist at $0. This workflow closes exactly that gap.
• The recording already happened. Zoom transcribes meetings going forward, cloud-recorded, on the right plan. A folder of last year's .m4a files can only be transcribed by something that takes uploads.
• You want the transcript out of the vendor's cloud. Here the uploaded file is deleted as soon as the transcript is ready and the transcript is deleted after 24 hours, with no account linking either to you — the privacy policy states it exactly, and the terms carry the limits.

The adjacent option is a meeting assistant (Otter, Fireflies) whose bot joins every call and archives it. That is a different product with a different trade-off — subscription, account, a bot participant visible to your counterparts — and the WhoSayAh vs Otter.ai comparison walks through when each shape wins. For meetings that cannot touch any external service at all — NDA, medical, legal, data-residency — the remaining route is self-hosted transcription inside your own network, which is FastoCloud territory: same upload-and-read workflow, no audio leaving your environment.

Consent: ask before you press Record

A Zoom recording is a recording of other people. In the EU, the UK, Canada and roughly half of US states, all-party consent is legally required; everywhere else it is still the professional norm. Zoom helps here — participants see a recording indicator and a consent prompt when recording starts — but do not lean on the indicator alone. Say it out loud at the top of the call: "I'm recording this for the minutes — any objection?" and stop if there is one. For formal contexts, note the agreement in the minutes themselves.

The same etiquette applies to what you do with the transcript: it quotes colleagues and clients verbatim. Share the minutes, not necessarily the raw transcript. If your calls happen on Google Meet instead, the Meet recording guide covers that platform's version of this whole workflow, including a free way to record when Meet itself will not.

Transcribe a Zoom recording