A recording that only captures a microphone gets one side of a call clearly and the other side secondhand — audio that travelled out of the speakers, bounced around the room, and went back in through the same microphone picking up your own voice. Recording both sides properly means two separate tracks instead: the microphone for your side, and the call's own audio, read directly, for the other one. Here's what that setup actually involves.
Why "both sides" means two tracks, not one
A single microphone recording mixes everything the room can hear into one file — your voice loud and close, the other side quieter and duller, run through whatever speaker played it first. Capturing the call's own audio separately avoids that entirely: it reads the audio stream the Mac is already producing, the same signal being sent to the speakers, rather than waiting for it to travel through the air and back into a microphone.
Step 1 — grant the one-time system audio permission
The first time a recording includes the call's own audio, macOS interrupts with a permission prompt rather than capturing it silently. That's a one-time step, not a per-call one: approve it once, and every later recording includes both tracks without asking again. The full detail on this permission covers what to do if it was dismissed the first time around.
Step 2 — start recording before either side is talking
Press record before joining the call, whether that's dialling into a meeting platform or answering a phone call through the Mac. Nothing said before recording started can be recovered afterward, which makes this the step most worth turning into a habit rather than remembering per call.
Step 3 — the microphone holds your side automatically
Nothing extra needs configuring here — the built-in or connected microphone captures your own voice the same way it always has, as its own track, completely separate from whatever the call itself is playing back.
Step 4 — system audio holds the other side, however the call arrives
Whether the call is a video meeting in a browser tab, a desktop app, or a phone call routed through the Mac, the same mechanism applies: system audio capture reads whatever the Mac is outputting at that moment. There's no setup specific to which platform the call happens to be running in.
Step 5 — confirm both tracks actually came through
Before relying on this for a call that matters, a short test — playing any audio through the Mac for thirty seconds and checking both the microphone and system audio came through — catches a permission that got silently reset by a macOS update before it costs an actual call's audio.
A phone call taken through the Mac, not a video meeting
The same two-track capture applies to a phone call answered on the Mac rather than a dedicated meeting platform — through a carrier's Wi-Fi calling feature routed to the Mac, or a call continued from an iPhone. System audio capture doesn't distinguish between a video meeting's audio and a phone call's audio; both are simply what the Mac is currently outputting.
A dial-in number opened in a browser tab
A conference line dialled into through a browser rather than a dedicated meeting app produces audio the same way any other call does, so it's captured identically — nothing about the call arriving as a phone number instead of a meeting link changes which tracks get recorded or how.
When "both sides" is more than two voices
A call with several other participants still produces one system audio track carrying everyone on the other end, rather than a separate track per person — the same as a single microphone in a room with several people present. What separates one voice from another afterward is matching each one against the others by how it sounds, not which track or line it arrived on.
What a recording missing the other side actually looks like
The most common version of "I can only hear myself" is the system audio track being silent while the microphone track is fine — usually because the one-time permission above was never granted, or got reset by a macOS update since the last time it was checked. That's worth ruling out before assuming anything about the call itself; the fix is the same permission check either way.
Speaker separation once both tracks exist
With both sides captured cleanly, the transcript that comes out attributes lines to different speakers by comparing how each voice sounds, building that comparison from the two tracks rather than guessing from which one a segment came in on. A label that starts generic can be corrected once and is recognised the same way in later calls.
Why one track isn't a workaround for the other
Turning the microphone up, or angling it toward the speakers, narrows the gap between the two sides without closing it — the other side is still a re-recording of a re-recording, with room noise and echo layered on top of whatever was actually said. Reading the audio stream directly, rather than capturing it acoustically a second time, is a different approach entirely, not a louder version of the same one.
Headphones don't change which side gets captured
Because system audio capture reads the audio stream itself rather than sound travelling through the air, wearing headphones during a call doesn't remove the other side from the recording the way it would if the setup depended on a microphone picking up what's coming out of the speakers. Both tracks are captured the same way whether the output is going to built-in speakers, headphones or an external interface.
No separate hardware or driver involved
Some methods for capturing a computer's own audio route it through a virtual audio driver or an external interface before anything can record it. None of that is necessary here — once the Screen Recording permission above is granted, both tracks are captured natively, with nothing extra to install or configure beyond that one approval.
The fuller process, once both tracks exist
Capturing both sides is the first half of a longer sequence, not the whole of it. The complete walkthrough for recording a meeting on a Mac covers what happens after the two tracks are captured — the live transcript, the speaker labels, and the summary written once the call ends — for anyone setting this up for the first time rather than fixing a single missing track.
Switching output devices partway through a call
Moving from built-in speakers to headphones mid-call, or connecting a different output device, doesn't interrupt system audio capture — it keeps reading whatever the Mac is currently outputting rather than staying tied to whichever device was active when recording started. The microphone track is unaffected either way, since it was never reading from the output side of the call to begin with.
Testing this on a call that doesn't matter first
Before trusting both-sides recording for a call where losing a track would actually cost something, a short test call — even a call to a voicemail system or a quick check-in with a colleague — confirms both tracks are coming through cleanly. That's a cheap way to catch a permission that was quietly reset by a macOS update before it costs an important call's audio.
Why this differs from just turning the volume up
Raising the Mac's own output volume before a call makes the other side easier to hear in the room, but it doesn't change what the microphone actually records — a louder secondhand copy is still a secondhand copy, picked up off the speakers with room noise and echo layered in. System audio capture reads the same signal the speakers are playing without depending on volume, distance from the microphone, or anything else about the room at all.
Keeping an archive of calls recorded this way
Calls recorded with both tracks intact accumulate the same way any recording does — one entry per call, staying on the Mac that made it rather than syncing anywhere else, across up to three machines on one licence. The wider case for keeping an entire archive of calls private covers what that looks like once there's a month of them to look back through, not just a single recording.
A call taken while away from a desk
Both tracks are captured the same way whether the Mac is sitting on a desk or being carried around during the call — a laptop's built-in microphone and its own audio output work identically for this purpose regardless of where the machine physically is, as long as it stays awake and the call stays connected through it. Nothing about the setup assumes a fixed location or a particular piece of external hardware plugged in.
Why this matters beyond just clarity
A recording missing the other side isn't only harder to listen back to — it's harder to transcribe accurately, since a transcript built from a muffled, secondhand copy of someone's voice has more to get wrong than one built from a clean signal read directly off the call. Two clean tracks aren't a nicety on top of the recording; they're what makes an accurate transcript, and the minutes built from it, possible in the first place.
What this looks like across a week of calls
Once both-sides recording becomes the default rather than something set up per call, a week of calls turns into a week of recordings where every one of them, not just the ones that seemed important in advance, has a clean copy of both sides available afterward. That matters most for the call that turns out to matter more than it seemed like it would going in — the one nobody thought to treat specially until after it had already happened.
A call recorded from a Mac that isn't hosting it
Being the host of a call isn't a requirement for capturing both sides of it — a Mac joining as one participant among several still hears its own microphone and plays the call's own audio through its speakers the same way, regardless of who scheduled the meeting or who else in the room is also recording it independently on their own machine.
A call that starts on one device and continues on another
Occasionally a call begins on a phone and finishes on a Mac, or the other way around, as a conversation moves from a quick mobile check-in to a proper call at a desk. Only the portion of the call actually happening through the Mac gets captured this way — there's no way to retroactively record the part that already happened somewhere else, which is worth knowing before assuming a recording covers a conversation from its very first word rather than from whenever it moved onto the Mac.
Where the transcript goes from here
Two clean tracks are the input; a transcript, built while the call still runs and never uploaded anywhere to produce it, is what comes out the other end. Neither track needs to leave the Mac at any point in that process for the transcript or the eventual summary to exist.
The short version, stated plainly
Both sides of a call means two tracks, not one louder microphone. Grant the system audio permission once, start recording before the call does, and the microphone and the call's own audio stay separate from the first second onward — the setup described above, done once, rather than a habit to repeat before every call that matters, and worth doing before the next call rather than after the one that turns out to have needed it most.
What it costs, what it runs on
Recording both sides of a call as two clean tracks is how Coii AudioNotes works by default — $19, paid once, for three of your own Macs, with a 30-day trial needing no card and no account. It runs on macOS 13 Ventura or later, Apple Silicon or Intel. Compared with Apple's own Voice Memos, which only ever captures a microphone, this is the specific gap that two-track recording closes.