Recording a meeting on a Mac, done properly, means capturing two things at once — the microphone in the room and the audio the meeting itself is playing through the speakers — as separate tracks, started before the call begins rather than fumbled into partway through. Everything below is that process broken into the steps that actually matter, using Coii AudioNotes as the tool doing the capturing.
What "recording a meeting" means here
There's no participant joining the call to do this. System audio capture means reading the audio the Mac is already playing — the other side of the conversation — the same underlying mechanism a screen recording uses, not a bot logging into the meeting under its own name. The microphone captures your own voice separately, and the two stay apart as distinct tracks rather than getting mixed into one file, which is what makes it possible to tell speakers apart afterward.
Step 1 — grant the permission once
The first time system audio capture is used, macOS asks for permission to let an app read what's currently playing. That's a one-time prompt, not a per-meeting one: approve it once in System Settings and every recording after that starts without the interruption. Skipping this step is the most common reason a first recording comes back with only one of the two tracks — the microphone captured, the call itself silent.
Step 2 — have the call already open
Nothing needs installing inside the meeting platform itself. Whatever you already use to join calls — a browser tab, a desktop app, doesn't matter which — stays exactly as it is. Coii AudioNotes doesn't integrate with a calendar or a meeting platform to do this, which also means there's no setup step per meeting beyond having the call open the normal way.
Step 3 — press record before you dial in
The habit worth building is starting the recording before joining the call, not partway through it once something worth keeping has already been said. There's no way to recover audio from before recording started — a limitation worth knowing up front rather than discovering it after losing the first five minutes of an important call once.
Step 4 — talk normally; transcription runs while the call does
Once recording starts, the transcript builds segment by segment as the conversation happens, rather than being generated afterward from the finished file. Nothing about how the meeting itself runs needs to change — no pause to let something catch up, no visible indicator on the call for anyone else to notice.
Step 5 — speakers get separated automatically
Each voice in the recording gets matched against the others using a voice print — a comparison of how each voice sounds, not an identity check — so the transcript comes out with lines attributed to different speakers rather than one unbroken block of text. Labels start generic and can be given a name once, after which the same voice is recognised the next time it shows up.
Step 6 — end the call, and the write-up finishes itself
When the meeting ends, the transcript is already complete; what happens next is the summary. A language model bundled with the app turns the transcript and any notes taken during the call into a short set of minutes — decisions, actions, open questions — generated on the Mac itself rather than sent anywhere to be written.
Which permission macOS actually asks for
The first time any app tries to capture system audio, macOS shows a specific one-time prompt tied to the Screen Recording permission category in System Settings → Privacy & Security — the same category the operating system uses for anything that reads what's currently on screen or playing, audio included. Approving it there is the entire setup; there's no separate account to create and no key to enter first. If a recording ever comes back with the microphone track present but the call's own audio missing, that permission not being granted is the first thing to check, ahead of anything more complicated.
It doesn't care which meeting platform is on screen
Because system audio capture reads whatever the Mac is already playing rather than integrating with any one meeting platform specifically, it works the same way whether the call is in Zoom, Google Meet, Microsoft Teams, a phone call routed through the Mac's own audio, or anything else that makes sound. There's no per-platform setup and nothing that breaks when a team switches which tool they schedule calls in.
What the transcript looks like while it's still happening
Rather than a blank screen until the call ends, the transcript fills in live, one segment at a time, attributed to a speaker as soon as there's enough audio to match it against the others already in the recording. That running text is there to read as the call goes on if it's useful — checking that a number got captured correctly, for instance — without needing to wait for the recording to finish to see anything at all.
Fixing a name a transcript got wrong
A voice print groups segments that sound like the same person before a name ever gets attached to it, which means a label sometimes starts generic — "Speaker 2" — until it's corrected once. That correction carries forward: once a voice has a name, later recordings featuring the same person recognise it again rather than starting over from a blank label each time.
If you forget to press record
It happens, and there's no way around it after the fact: audio that was never captured can't be recovered from a transcript that was never started. The nearest comparison is Apple's own Voice Memos, which only ever captures the microphone in the room and has no concept of the call's own audio at all — Coii AudioNotes against Voice Memos covers that gap in more detail. Either way, the fix for a forgotten recording is the same going forward: start it as part of joining the call, not as an afterthought once the conversation is already interesting.
Recording more than one call in a day
Nothing about finishing one recording changes the steps for the next — press record again before the next call starts, the same as the first. Each recording, transcript and set of minutes stays a separate file rather than getting merged into a running log, which matters once there's a week's worth of them to search back through later.
Keeping the Mac awake for the length of the call
Recording depends on the Mac staying awake for as long as the meeting runs, the same requirement any app capturing audio in the background has — the ordinary energy-saver settings that keep a laptop from sleeping mid-call apply here exactly as they would for the video call itself. A Mac that's plugged in with its screen open for the length of a normal meeting doesn't run into this at all; it's mainly worth knowing before recording an unusually long session on battery with the lid nearly closed.
More than one person talking into the same microphone
A recording made from one Mac in a room with several people present still separates by voice rather than by device — voice print matching works from the audio itself, not from which laptop or microphone captured a given line, so an in-person meeting with everyone around one Mac gets speaker labels the same way a video call with everyone on a separate device does.
Exporting a finished recording
Once a call is transcribed and the minutes are generated, the recording and its transcript are files that can be moved, copied or handed to whatever backup routine already covers the rest of the Mac — there's no separate export process to learn because there was never an upload step to begin with.
What this workflow doesn't include
There's no bot to schedule onto a calendar in advance, no per-meeting setup inside Zoom, Meet or Teams, and no dashboard anyone else logs into to see the result — recording, transcribing and writing minutes all happen on the one Mac that's actually in the room or on the call, for the person who pressed record. A team that wants a shared workspace where several people browse a library of past meetings from their own login is describing something this workflow doesn't do.
Keeping the archive somewhere sensible
Recordings accumulate quickly once this becomes a habit — a week of client calls is a week of files. They stay on the Mac that made them rather than syncing to an account, across up to three machines on one licence, which means the ordinary rule for anything on a Mac worth keeping applies: back the folder up the way you would anything else. The fuller case for keeping an entire meeting archive private covers what that looks like across a larger set of recordings, not just one call.
Recording an in-person meeting, not just a call
None of this requires the meeting to be happening over a video platform at all. A board meeting in a conference room, a client sitting across the table, an interview conducted face to face — any of them can be recorded the same way, just with the microphone doing most of the work rather than system audio, since there's no call audio for the Mac to be playing in the first place. Speaker separation still works the same way afterward, built from how each voice sounds rather than from which side of a call it came in on.
What a finished recording actually contains
By the time a call ends, what exists isn't a single audio file waiting to be processed later — it's a recording, a transcript that was largely finished as the call happened, and shortly after, a set of minutes, bundled together as one entry in the archive rather than three things a person has to assemble by hand. Opening that entry later shows all three at once: the original audio to play back if needed, the transcript to scan, and the summary to read first.
Finding a specific meeting again later
An archive that grows by one entry per call becomes hard to navigate by memory alone after a few weeks of steady use. Being able to come back to a specific meeting — the one where a particular number got discussed, or a particular decision got made — depends on more than remembering the date; it depends on being able to search across what was actually said rather than scrolling through file names. That's a separate workflow from the recording step itself, worth treating as its own habit once there's more than a handful of recordings to look back through.
When more than one Mac is in the same room
Only one Mac needs to be recording for an in-person meeting to be captured — a second laptop open in the same room isn't a second recording that needs merging with the first, it's just a second laptop. The person running the meeting, or whoever is designated to keep the record, is the one who presses record; everyone else can leave their own Mac out of it entirely without the transcript missing anything they said, since a single microphone in the room picks up the whole conversation.
The difference between recording and joining
It's worth restating plainly, because the distinction is easy to lose sight of once the steps above become routine: nothing here ever logs into the meeting as a participant, requests access to a calendar, or appears on a roster anyone else can see. Recording is something that happens on one Mac, quietly, alongside a call that looks, to everyone else on it, exactly like a call with no recording happening at all.
What it costs, what it runs on
Coii AudioNotes is $19, paid once, covering three of your own Macs, with a 30-day trial that needs no card and no account to start. It runs on macOS 13 Ventura or later, on Apple Silicon or Intel — there's no separate app to install for system audio capture specifically, and no account to create before the first recording. A fuller look at what the app is built to do covers the rest of it beyond this one workflow.