Meetings on Mac

Record a call or an in-person meeting, get a transcript with speakers, a summary and action items, and play the audio back. All processed on the Mac.

Mac only.

Start a recording

Choose Record meeting from the menu bar during a call, or click Record (⇧⌘R) in the Library toolbar.

A sheet says Ask everyone for permission before you record. Recording laws vary by country. It also offers Tell speakers apart, a one-time 13.7 MB model download that labels who said what. Tick Don’t show again if you don’t need the reminder, then click Start recording.

What you’ll see: the menu bar icon gains a small coral dot, the menu item becomes Stop recording · 12:04, and the Library shows the timer with separate Microphone and Call audio levels.

What is captured

Your microphone, labelled You.
The other side: the audio of the apps on the call, through macOS’s system audio capture. The first time, macOS asks for the System Audio Recording Only permission.

In-person meetings work too: the Call audio row reads None — in-person, and the speaker model tells people apart from the microphone alone.

Dictation keeps working during a recording; the floating bar just stays hidden.

While recording

Pause and Resume let you take a break; Stop finishes and starts processing. Capture is saved as it goes, so if Scribe quits unexpectedly it recovers the recording on the next launch.

After you stop

Processing shows four steps: Saving audio, Transcribing, Identifying speakers (when the model is installed) and Summarising. Scribe transcribes both streams, labels turns as You and Speaker 1, Speaker 2 and so on, and writes a summary of up to six bullets plus a checklist of action items with owners where they were said.

Summaries use your Refine engine: Gemma 4 summarises the whole transcript in one pass; Apple Intelligence works in chunks. Very short meetings show Too short to summarise.

Open the meeting in the Library window (Open Scribe…). Meetings sit below your notes in the sidebar, grouped by day.

Click a speaker name to rename it; the change applies to all their turns.
Add your own notes; they save automatically. Double-click a transcript turn to edit it.
Play the audio, if it was kept, and tap any timestamp to jump there. See Meeting audio playback.
Copy offers Summary, Transcript or Everything. Export as Markdown or plain text.
Update summary regenerates the summary after you edit the transcript; Retry transcription reprocesses the audio, for instance after installing the speaker model.

Settings › Meetings

Show reminder before recording: the consent sheet.
Keep recordings’ audio: off by default. Scribe keeps the transcript and deletes the audio after processing unless this is on.
Speaker detection · 13.7 MB: download or delete the speaker model.

Privacy

Meeting audio is processed on your Mac and never sent anywhere. Transcripts, summaries and any kept audio are stored in Application Support alongside your notes. Delete a meeting to remove all of it; it waits 30 days in Recently Deleted.

Common problems

No call audio. Check System Settings › Privacy & Security › Screen & System Audio Recording and make sure NanthAi Scribe is allowed. The Call audio row says Not coming through while nothing arrives.

“The meeting language isn’t installed.” Transcription uses your dictation languages. Add the language spoken in the meeting under Settings › Languages.

Everyone is “Speaker 1”. Install the speaker model under Settings › Meetings, then choose Retry transcription in the meeting’s menu.

The summary is empty or wrong. Edit the transcript, then Update summary. If Refine isn’t available, see Refine.