Home / Blog / How to transcribe a meeting recording and know who said what

How-to

How to transcribe a meeting recording and know who said what

Turn a Zoom, Teams or Meet recording into a transcript with speaker labels, timestamps and a PDF, in a few minutes and without installing anything.

October 9, 2026 · 4 min read

You have a recording of a meeting and you need what was said in writing: the decisions, the action items, the exact wording of a promise. Listening back at double speed still takes half the meeting's length. This is the shortest path from the recording to a transcript you can search, quote and send, with each speaker labelled.

1. Find the recording file

• Zoom saves local recordings under Documents → Zoom, one folder per meeting, with an audio_only.m4a next to the .mp4. Either one works; the audio file uploads faster. Cloud recordings can be downloaded from the Zoom web app under Recordings.
• Microsoft Teams puts recordings in OneDrive or SharePoint, in a Recordings folder. Download the .mp4.
• Google Meet saves to Google Drive in a Meet Recordings folder. Open the file in Drive and choose Download; a Drive share link will not work, because it points at a web page rather than the file.
• Phone voice memos export as .m4a from the share sheet.

Any common audio or video format is fine: MP3, M4A, WAV, MP4, MOV, WebM and more, up to 5 GB. A video is handled like audio; only the sound is used.

2. Create a free account

Go to audiotranscribr.com/app and choose Create a free account. Enter your email address and the one-time code we send. There is no password to invent and no card to enter. Your first transcription is free whatever its length, so a two-hour meeting is fine.

3. Upload the file

On the Transcriptions tab, drag the recording onto the page, or click choose a file. A progress bar shows the upload; a one-hour meeting is usually ready about a minute after the upload finishes. Long video files take a few minutes longer because the audio track is extracted first.

4. Download the transcript

When the status turns to completed, five download buttons appear:

What speaker labels look like

Speakers are told apart by voice and labelled Speaker 1, Speaker 2 and so on, in the order they first talk:

[00:00:07] Speaker 2: Mister chief justice, and may it please the court, this case well illustrates the real world cost of Chevron …

The service does not know people's names, so the last step is yours: open the TXT file, find the first line each speaker says, and replace Speaker 1 with the person's name throughout. In most editors that is one find-and-replace per speaker.

Two things help the labels come out clean:

• One voice per microphone. A laptop in the middle of a conference room gives everyone the same acoustics, which is the hardest case. Headsets or a meeting app that records each participant separately give the best results.
• Let people finish. Overlapping speech is where any transcription, human or automated, loses the thread.

Privacy

The recording is deleted from our storage seven days after upload and the transcript after thirty, or immediately if you delete the transcription yourself. Audio is used only to produce your transcript; it is never used to train anything. The privacy policy has the full version.

If you do this every week

Everything above is also available as an API, so a script or an automation tool like n8n can send every new recording for transcription and file the transcript where your team already looks for it.

Frequently asked questions

Keep reading