Bot-free meeting notes

Why a meeting recorder should keep you and them on separate tracks

Recording your microphone and your computer's audio as two files lets a transcript tell your side from theirs without voiceprints. How, and the pitfall.

By the Notey team at AInject · · 6 min read

In short

A two-track meeting recording keeps your microphone and what your computer plays as two separate files. Your voice is on one; everyone else on the call is on the other. That single decision is how a transcript can label each line "You" or "Them" reliably, without learning anyone's voice.

A single mixed recording loses that information the moment the two are combined. This guide explains why separate tracks matter, what they can and cannot tell you, the one situation where they go wrong — speakers — and how to fix it.

One mix, or two tracks

On a call, your computer handles two streams of sound:

  • The microphone input, which carries your voice to the call.
  • The audio output, which plays the other people to you.

Most ways of recording a call merge the two. A screen recorder with audio, a virtual audio device that mixes inputs, or a phone held near the laptop all produce one track with everybody on it. To find out who said what, software then has to work it out from the sound of the voices alone.

A two-track recorder captures the input and the output separately and keeps them separate. Nothing has to be worked out about which side spoke: the file it came from says so.

One mixed trackTwo tracks
Your side vs theirsGuessed from voicesKnown from the source
Crosstalk (two people at once)Words from both run togetherEach side transcribed on its own
Voice recognition needed to label sidesYesNo
Several people on the other sideNeeds diarisationStill needs diarisation on their track
Echo from speakersInvisible, already mixedShows up as them on your track

How to record your microphone and your Mac's audio at the same time compares the ways to set this up on a Mac.

What separate tracks give you

Your side and theirs, without voiceprints

Telling voices apart from sound alone is called speaker diarisation. It works reasonably well, but it guesses — it can merge two similar voices or split one person into two when they change tone. With two tracks, the split between you and the others is not a guess at all. No voice has to be learned, stored or compared, which also matters if you would rather not keep biometric data about the people you talk to.

Crosstalk that stays readable

When two people talk at once on a mixed recording, the recogniser hears one tangle of sound and often drops words from both. On two tracks, your words and theirs are each clean on their own file, so both are transcribed.

A check that each side was actually captured

With one track, you cannot tell from the file whether the other side was recorded or only you were. With two, a silent track stands out. A recorder can notice when one side stops arriving and say so, rather than hand you half a meeting as if it were whole.

A cleaner input for summaries

Knowing which lines are yours matters for notes: action items you took on, questions you asked, where you and the other side disagreed. That is only as reliable as the labels underneath it.

What separate tracks cannot do

Two tracks tell you which side spoke, not which person on that side.

  • Several people on the other side all arrive on the same track. Telling them apart is diarisation again, applied to their track only. Speaker diarisation, explained covers how that works.
  • Several people in the room with you all arrive on your microphone. In an in-person meeting, the two-track split gives you nothing to go on.

Where it goes wrong: speakers

Two-track labelling assumes your microphone hears only you. That holds with headphones. It does not hold on speakers.

On speakers, the other people's voices come out of the laptop and straight back into its microphone. The call app's echo cancellation removes that from what the others hear, but a recorder capturing your raw microphone still picks it up. The result is their words appearing on your track — sometimes twice in the transcript, once on each side, sometimes only on yours.

The fix is headphones. Any wired or wireless pair keeps the other side's audio out of the microphone. Fixing echo in call recordings covers the other causes and remedies.

If you record on speakers anyway, expect to correct a few lines afterwards.

How to tell whether a tool records two tracks

Vendors describe this in different words. Look for:

  • A transcript that labels "Me" / "Them" or "You" / "Them" without you naming anyone, on every app you use.
  • Separate level meters for your microphone and the other side while it records.
  • An export that includes separate audio files, or a setting for them.

Zoom's own local recording can save a separate audio file per participant; that is a platform feature for Zoom meetings only, and it works differently from recording at the computer.

How Notey records two tracks

Notey, a Mac app for Apple silicon and macOS 26 or later, records every meeting as two tracks: your microphone, and what your Mac plays. It is one of the ways to take bot-free meeting notes: nothing joins the call.

  • You and Them come from the track each line arrived on. You can give either side a name for that meeting.
  • Two meters on the floating button, labelled You and Them, show each side separately while it records.
  • On speakers, the other people can end up on your side. If it happens, you can move a line to the other side; headphones stop it happening.
  • Several people on the other side are told apart in a pass after the recording stops. You name them, and mark where somebody starts speaking.
  • If one side stops arriving, the meeting is marked incomplete and says why, instead of looking complete.
  • Voice recognition across meetings is separate, off by default, and only offers suggestions you accept or reject.

Both tracks are transcribed on the Mac with Apple's speech recognition and stay on your disk.

Get a good two-track recording

  1. Wear headphones. This is most of it.
  2. Check the input. Make sure the microphone the call uses is the one you are recording. Headsets and docks can change it.
  3. Pick a quiet spot. Background talk near your microphone lands on your track and can become words in the transcript.
  4. Do a ten-second test before an important call: talk, play a video, and check both sides have sound.

Frequently asked questions

What is two-track call recording?

Recording your microphone and the audio your computer plays as two separate files instead of one mixed file. The first holds your voice; the second holds everyone else on the call.

Why does my voice appear on the other side of the transcript, or theirs on mine?

Usually because you are on speakers. The microphone hears the other people through the speakers and records them on your track. Headphones stop it.

Does two-track recording need voice recognition?

No. It knows your side from theirs by which input the sound came from. Telling apart several people on the other side is a separate step, called speaker diarisation.

Can I record two tracks with QuickTime?

QuickTime Player records one audio input at a time. Recording the microphone and the Mac's audio separately needs a tool built for it, or a more involved setup with virtual audio devices.