Who spoke when?
The audio is cut into turns, and each voice gets a neutral label. Meetlens does this automatically on every recording.
Meetlens transcribes your Zoom, Google Meet, Teams and Webex calls, or a file you upload, and splits the text by speaker, every line with a timestamp. Rename Speaker B to Nora once, and the whole meeting follows.
Every meeting starts with neutral labels: Speaker A, Speaker B, Speaker C. Give one a name and the transcript, the timeline and the AI notes all switch to it.
Speaker ASpeaker BNoraSpeaker C
Thanks for joining, Nora. Shall we start with pricing?
Sure. We'd like to keep the monthly plan.
We can, if we drop the setup fee instead.
Then finance has to sign off on the discount.
Fine. I'll send the revised quote by Thursday.
Send the revised quote
Speaker separation runs on every recording, on every plan. There is nothing to switch on.
Record the call from Chrome or the Mac app, or upload an audio or video file.
After the call, the transcript is split into turns, each labelled Speaker A, B or C, with a timestamp.
Replay a line if you're unsure who said it, then put real names on the labels.
Part of every meeting in Meetlens: Meeting transcription →AI meeting notes →Meeting recording →
The two get mixed up. Meetlens does the first on its own and leaves the second to you.
The audio is cut into turns, and each voice gets a neutral label. Meetlens does this automatically on every recording.
A real name goes on the voice. In Meetlens, you type it or confirm a suggested one. There are no voiceprints: voices are told apart inside one recording, never matched across meetings.
Separating voices is an automatic draft. These are the things that make it harder, and what helps.
In our own test on public meeting recordings, Meetlens found all four speakers in every excerpt. Most wrong labels were short replies like “yeah”; on turns of five words or more, 2% to 6% carried the wrong label.
How accurate is AI transcription? Our test →Meetlens records from your own computer, so no bot joins the call. Tell people before you record.
On Windows, the Zoom and Teams apps can't be recorded: join the same meeting in your browser.
Upload an MP3, WAV, M4A, OGG, WebM or MP4 recording, up to 1 GB (3 GB on Business), and get the same speaker labels and timestamps. Handy for a meeting in a room: record it on any device, then upload it. A file can run 60 minutes on Free, 90 on Pro and 120 on Business.
Transcribe a recording you already have →
Everything above is on Free. File export (PDF, Markdown, TXT) is on Pro and Business.
Compare plans →“It doesn't just give you a transcript, it pulls out what actually matters: decisions, who's doing what, next steps. Best part is I can just focus on the conversation instead of typing notes the whole time.”Chrome Web Store review
“Used to spend like 30 minutes on a summary after each call, now it's 5. My team actually reads the summaries now instead of ignoring them.”Chrome Web Store review
“We also mix languages quite a bit during calls, and it still gives us clear reports in the language we've selected in the settings.”Chrome Web Store review
Rating, users and reviews from the Meetlens listing in the Chrome Web Store, October 2026. Reviews are shortened.
Transcription · start hereHow Accurate Is AI Transcription?How accurate is AI transcription? What word error rate means, what research shows, and our test of Meetlens on real meeting recordings.Our meeting testCrosstalkSpeaker labelsBetter audioWith Meetlens you don't have to mark them by hand. After the call, the transcript is split into turns labelled Speaker A, Speaker B, Speaker C, each with a timestamp. Open Manage speakers, type a name next to each label and press Save all: the transcript, the timeline and the AI notes switch to the names.
Speaker diarization means working out who spoke when: the audio is cut into turns and each voice gets a neutral label. It doesn't tell you who the person is. Meetlens runs diarization on every recording; putting a real name on a label is a separate step that you do or confirm.
Yes. Meetlens labels speakers on the Free plan too: 180 minutes of transcription a month, up to 60 minutes per recording, with speaker labels you can rename and timestamps on every line. No credit card is needed.
Record the call with the Chrome extension or the Mac app, or upload a recording: MP3, WAV, M4A, OGG, WebM or MP4, up to 1 GB (3 GB on Business). Meetlens returns a transcript split by speaker, with a timestamp on each line. For a meeting in a room, record it on any device and upload the file.
No. Meetlens creates no voiceprints. It tells voices apart inside one recording only and never matches them across meetings, so every new meeting starts with Speaker A, B, C. If someone is called by name during the call, Meetlens may suggest that name; nothing changes until you apply it.
It depends on the audio. One person speaking at a time and a close microphone give the cleanest result. Crosstalk, one shared microphone for a whole room and similar voices make labels slip, and a one-word reply such as “yeah” can land in the neighbouring speaker's line. Check important quotes against the recording before you share them.
Yes. Once you rename a speaker, the transcript, the timeline and the AI notes show the name, and tasks list that person as the owner. You can rename again at any time.
Yes. Upload the file and it goes through the same transcription, with the same speaker labels and timestamps. A file can be up to 60 minutes long on Free, 90 on Pro and 120 on Business.
Use the Chrome extension or the macOS app. 180 free minutes a month, no credit card.