Speaker names
How voices become names, from telling voices apart on your Mac and reading the meeting window, and how to fix a name yourself.
How voices get names
Your microphone is a separate channel, so anything you say is yours for certain, and the transcript labels it You. Everything else arrives through system audio, and Cluinn tells those voices apart on your Mac so each person keeps their own line.
What each line is called depends on how much is known:
| Label | What it means |
|---|---|
| You | Your own microphone |
| A name | Cluinn read this person’s name from the meeting app |
| Speaker 1, Speaker 2, … | A distinct voice or tile it has told apart but not named |
| Unknown speaker | A line it couldn’t tie to any particular voice |
Names come from what the meeting app itself shows. The first source needs no Screen Recording: with the Accessibility permission Cluinn asks for at setup, it reads the text an app makes available to macOS, which is usually where the participant list is. Decline Accessibility and that source is gone. With the Screen Recording permission (called Screen & System Audio Recording on macOS 15 and later), Cluinn can also read the window as a picture: the names on the tiles, and which tile is highlighted as speaking. A name is written only when the sources agree; when they disagree, Cluinn leaves the name out rather than guess.
It looks at that one window, never your whole screen, and if the meeting window can’t be isolated, nothing is captured at all. No picture is kept: each one is deleted as soon as it’s read, and nothing from it goes into the saved call. Some of that reading happens on your Mac and some on Cluinn’s servers. It isn’t a single snapshot either: while people are still joining, the window is read every 20 seconds or so, with an extra look whenever a new voice starts, and once the list of names stops changing it’s read much less often. No audio is ever sent anywhere for naming, and no voiceprint of anyone is ever saved: voices are told apart during the call only.
A name that was confirmed on screen can carry over into a screen share, when the tiles are hidden. When Cluinn isn’t sure, you see Speaker 1 rather than the wrong name.
Fix names yourself
People lists everyone who spoke. The same button opens it in three places:
| Where | Who it lists | What a change does |
|---|---|---|
| A saved call: top right, next to Transcript, or at the top of the transcript | That call’s speakers | Changes that call only |
| A folder: top right | Everyone who spoke on the folder’s calls | Renames or merges the person in every call |
| All your calls: top right | Everyone who spoke on any call | Renames or merges the person in every call |
Each row has a colour dot, the person’s name and a name box. In a call, the name is followed by how many lines that person has, and the box of a voice with no name yet reads Name this speaker. In a folder or on All your calls, it’s followed by how many calls they’re in. Click the name box to pick someone from the list (picking an existing person merges the two) or type a new name. Nothing changes until you press Save. The window stays open after saving, with Undo next to Save, and ⌘Z works too once you’ve closed it. In a long list, the search field stays at the top while the rows scroll.
The dot sets that person’s colour, and it follows them to every call. No two people on one call share a colour, and up to 8 people get colours everyone can tell apart, colour-blind readers included. On bigger calls some colours can look alike, which is why the name always sits next to the line.
In a folder or on All your calls, Looks like the same person lists two names that seem to be one person spelled two ways, such as “Priya Sha → Priya Shah”. Merge keeps the name on the right (the one used in more calls), and Not the same stops suggesting the pair. Cluinn never merges two names without you.
Reassign speaker on a single turn moves that block to someone else: type a name and press ↵, or click one of the call’s other speakers listed under the field.
Edits happen on your Mac, cost nothing and are available on every plan. Renames, merges, colours and Not the same answers are saved into your calls, so where encrypted backup is switched on for your account, they reach your other Macs with them.
Note: Speakers can’t be edited while a call is still finishing transcription. Give it a moment and try again.
Common questions
Why do I see “Speaker 1” instead of a name?
Names come from what the meeting app shows: the text it makes available to macOS, plus, with Screen Recording, the tiles themselves and who is highlighted as speaking. If none of that names the person, or the sources disagree, the voice stays Speaker 1 rather than being guessed. A missing name is easy to fix with People; a wrong one is easy to miss. Rename the speaker afterwards and the whole call updates.
Screen Recording is still what Cluinn needs to hear the other side of the call, so it’s worth granting either way, but a name can arrive without it.
Why is one line “Unknown speaker”?
That line couldn’t be matched to any of the voices in the call, usually because it was a short interjection or people talked over each other. Use Reassign speaker on it to place it yourself.