Google Meet transcript with speaker names

A Google Meet transcript with speaker names tells you not just what was said but who said it, line by line. MeetFlovv builds one for every meeting it captures: each segment has a timestamp, the text, and usually the name of the person speaking, taken from the speaker timing in Google Meet's captions. When MeetFlovv cannot single out one speaker, it leaves the name blank.

This article explains how the names get there, why some lines have none, and how to get the most accurate result.

Why speaker names matter

A transcript without names is a wall of text. You can see that someone agreed to send the proposal, but not who. You can see an objection, but not whose. For notes and follow-up, the name is often the most important word on the line.

Names also help the AI notes: when a line is attributed to Sam, the notes can say Sam raised or agreed to something.

Two separate jobs: words and names

MeetFlovv treats "what was said" and "who said it" as two separate problems, solved by two separate sources. Speaker identification does not rely on recognizing voices.

The words come from the audio. The meeting audio is sent in 15-second pieces and each piece is transcribed by a speech-to-text service into short segments, each with a start and end time.

The names come from Google Meet itself. While captions run, Google Meet shows who is speaking. Both the Chrome extension and MeetFlovv Bot record that speaker timing, for example "Sam spoke from 1.0 s to 5.2 s", and send it to MeetFlovv alongside the audio.

How a line gets its name

For each transcript segment, MeetFlovv looks at the speaker timing and starts with one question: whose speaking time overlaps this segment the most?

• If one person clearly overlaps it the most, the segment gets that person's name.
• If nobody's speaking time overlaps it, MeetFlovv uses the speaker whose turn is closest, within about two seconds, but only when exactly one speaker is closest; otherwise the name stays blank.
• If two people overlap it by exactly the same amount, the name stays blank too.

That last rule is deliberate. A blank name is honest: you know to check. A wrong name is worse than none, because it looks right and sends an action item to the wrong person.

What the transcript looks like

On the meeting page in your dashboard, the transcript is a list of segments in order. Each one shows:

• a timestamp from the start of the meeting, such as 12:40,
• the speaker's name, when MeetFlovv could match one,
• the text of what was said.

The transcript fills in while the meeting is still running, shortly behind the conversation, because audio arrives in 15-second pieces. That is close enough to catch up if you join late. The notes, with the summary, decisions and action items, are written after the meeting ends. AI meeting notes for Google Meet, explained covers that part.

You can export the transcript as Markdown or plain text, with the timestamps and names included.

Why a line might have no name

A few situations leave gaps:

1. People talking over each other. If two voices share a moment equally, MeetFlovv does not pick one.
2. Very short interjections. A quick "yes" may sit more than a couple of seconds from any caption timing.
3. Captions not running. Speaker timing comes from Google Meet's captions. If captions were switched off during the meeting, there may be no timing to match against.
4. Audio from a shared room. If several people speak from one laptop in a meeting room, Google Meet sees one participant, so the transcript shows that participant's name for all of them.

None of these affect the words themselves. The text is still there; only the name is missing.

Getting the most accurate names

• Keep captions on in Google Meet. MeetFlovv tries to turn them on for you when capture starts; they give it the speaker timing it needs, so leave them on.
• Use your own device when you can. One person per participant tile means one name per voice.
• Avoid talking over each other. Good for the meeting, and good for the transcript.
• Set a display name in Google Meet. The transcript uses the name Google Meet shows, so "Sam Lee" is more useful than "Conference Room 3".

Speaker names in the notes

Names in the transcript carry through to the notes in one specific way. When someone says who will do something, such as "Sam will update the budget by Friday", the action item gets Sam as its owner and Friday as its due date. When a task is mentioned without an owner, the action item has no owner. The AI model is told never to invent owners or deadlines, and the speaker names do not change that rule: the person who mentions a task is not assumed to be the person doing it.

The summary and decisions can refer to people by name too, but only as the transcript supports. If a line has no speaker name, the notes do not guess one for it.

Extension or bot: does it change the names?

No. Both capture methods send the same kind of speaker timing, and the matching works the same way. The choice between them is about how you want to capture the call, not about transcript quality. Google Meet transcription: extension or bot? compares the two.

How long the transcript is kept

The transcript, including the speaker timing, is kept for 48 hours after your notes are ready, and then deleted. The notes themselves stay until you delete them. If you need the full meeting transcript for longer, export it before it expires. Meeting recording privacy: where your audio goes explains the full timeline.

Try it

Capture your next Google Meet call with the Chrome extension or MeetFlovv Bot, leave captions on, and you will have a transcript with names on the meeting page while the call is still going. See the features page for everything else MeetFlovv does.