Skip to content

Fix

One side of the conversation is missing from the transcript because one recording track didn't capture.

A transcript showing only one speaker means one track didn't record. Check System Settings permissions, restart app, and verify both tracks.

· 10 min read

On this page (29)
  1. What should be recording into each track
  2. The most common cause: system audio capture didn't start
  3. Checking the permission first
  4. The difference between permission denied and permission toggled off
  5. Restarting the app after granting the permission
  6. The silent track itself isn't broken
  7. Recording from a different app or tab mid-call
  8. Both sides missing vs one side missing
  9. Volume discrepancy between the two tracks
  10. What the waveform should look like
  11. The transcript starting empty vs starting with only one speaker
  12. Recording an interview where both sides aren't balanced
  13. How to visually confirm both tracks
  14. Testing the microphone separately
  15. Permissions needed for both tracks
  16. App focus and background suspension
  17. The importance of system audio for calls
  18. Why system audio capture is more prone to issues
  19. Conversation balance and speaker identification
  20. Recording calls from different apps
  21. How to tell if the issue is permissions versus something else
  22. The next recording after a failed one
  23. Moving forward after checking permissions
  24. Comparing to other recording scenarios
  25. Troubleshooting when permissions don't solve it
  26. Why both tracks matter for the full picture
  27. When one track failing means the whole thing fails
  28. The permission system protects everyone
  29. What it costs, what it runs on

A transcript showing only one person speaking — you, or the other person, but not both — means the recording only captured one of the two tracks it's supposed to keep separate: either the microphone or the system audio, but not both. One track went silent and the transcript can only show what it actually recorded.

What should be recording into each track

Coii AudioNotes records two things at once: the microphone in the room (your own voice) and the audio the Mac is playing (the other side of the call). These are two separate tracks, kept apart so a voice print can tell them apart afterward. If one track is missing, one side of the conversation never made it into the file at all.

The most common cause: system audio capture didn't start

The microphone usually records fine on its own. The missing voice is almost always the other side of the call — which means system audio capture didn't work. This happens for two reasons: the permission wasn't granted, or the app restarted mid-call and lost the capture session.

2 speakers
  • 00:02:14youWe should ship before the price change takes effect.
  • 00:02:23MayaRight, but we're missing a second release of the importer.
Both speakers present, identified by voice print, with timestamps. A missing voice means one track never recorded.

Checking the permission first

Open System Settings → Privacy & Security → Screen Recording. Find Coii AudioNotes in the list. If it's not there, the permission has never been requested. If it's there but toggled off, it was denied. Toggle it on and try recording again. If it's already on and the last recording still came back with one side missing, the issue is something else.

The difference between permission denied and permission toggled off

Both prevent system audio capture: denied at the prompt means the permission was asked for and rejected; toggled off in System Settings means it was granted before but then disabled. Either way, toggling it on in System Settings fixes it for future recordings.

Restarting the app after granting the permission

If you just granted the permission, close the app completely and restart it. The permission takes effect for new instances of the app, so an app that was already running when the permission was granted might not pick it up until it's restarted.

The silent track itself isn't broken

A missing voice isn't the same as a broken recorder. If both tracks recorded, one of them would have audio even if the call itself was quiet. A completely silent track means the system didn't attempt to record it at all — either the permission wasn't there or the source wasn't available.

Recording from a different app or tab mid-call

If the Mac is playing the call audio through a different app or browser tab than the one where it started, and the recording didn't follow it, one track might stop capturing. The microphone still works because it's always on, but system audio capture might lose track of where the sound is coming from. Closing and reopening the app won't help here — the fix is keeping the call in the same place for the entire recording.

Both sides missing vs one side missing

If the entire recording is silent — no microphone, no system audio — that's a different problem from one voice missing. One missing voice means one track worked and one didn't. No voices at all means something prevented recording from starting at all.

Volume discrepancy between the two tracks

The microphone and system audio are captured from different sources and naturally have different volume levels. A quiet microphone and loud system audio (or vice versa) is normal and doesn't mean one track failed — check that both tracks have any waveform visible in the recording, even if quiet. A completely flat track means it didn't record.

What the waveform should look like

When both tracks are recording correctly, the finished recording shows two separate waveforms, one above the other — one for the microphone, one for system audio. Each should show movement and variation. A completely flat waveform means that track captured nothing.

The transcript starting empty vs starting with only one speaker

If the transcript starts filling in live during the call but only shows one speaker from the start, system audio capture didn't engage from the beginning. If it was working and then stopped mid-call, the app might have restarted or lost focus. Keeping the app in focus for the entire call helps prevent it from being suspended mid-recording.

Recording an interview where both sides aren't balanced

If you're interviewing someone and your own voice is much quieter because you're just asking questions while they're doing most of the talking, both tracks are still recording — one is just quieter than the other. A missing voice is different: one track has no waveform at all. The presence of any visible waveform, regardless of volume, indicates the track was actually captured. Silence with zero visual indication means the track was never recorded.

How to visually confirm both tracks

Before troubleshooting further, open the completed recording and look at the waveform display. If both tracks are present, you'll see two distinct areas — one for each track. Each should show some visual indication of audio being recorded, even if very quiet. A completely blank area indicates that track never recorded. This visual confirmation is the fastest way to determine whether you have a permission issue versus something else entirely.

Testing the microphone separately

If only the system audio is missing but the microphone is clearly there, the problem is specifically with system audio capture — check the permission in System Settings under Privacy & Security → Screen Recording. If the microphone is missing but system audio is there, something prevented the microphone from capturing, which is rare and usually means the app wasn't given permission to access the microphone itself. Check System Settings → Privacy & Security → Microphone.

Permissions needed for both tracks

System audio capture requires Screen Recording permission. Microphone capture requires Microphone permission. Both are needed for a complete recording. A recording with only system audio and no microphone means the Microphone permission isn't granted. Check both permission categories in System Settings and ensure the app has both toggles enabled.

App focus and background suspension

macOS can suspend apps that run in the background to save resources. If an app loses focus or gets suspended during recording, it might stop capturing. Keeping Coii AudioNotes visible and active during the entire recording can help prevent this. The app needs CPU resources to capture both tracks simultaneously, and background suspension can interrupt this.

The importance of system audio for calls

When recording a video call through Zoom, Google Meet, Microsoft Teams, or any other platform, the audio from that call is routed through the Mac's speakers and captured by the system audio track. The microphone captures your own voice and any ambient sound in your room. Together, these create a complete record. Without system audio, you lose the entire context of what the other person said.

Why system audio capture is more prone to issues

The microphone usually captures reliably because it's a local device attached to or built into the Mac. System audio capture is a more complex operation that requires both the permission and the app to actively listen to what the Mac is playing. This is why system audio is more commonly the source of missing audio — it has more moving parts and more points of failure.

Conversation balance and speaker identification

When both tracks are present, the voice print system that identifies speakers works well because it has distinct audio to compare across the two sources. It can identify your voice from the microphone track and match it against the other speaker's voice in the system audio track. If one track is completely missing, speaker identification becomes impossible.

Recording calls from different apps

Different video conferencing apps — Zoom, Google Meet, Teams, Webex — all route their audio through the Mac's speaker output, which is why system audio capture works regardless of which app you're using. However, if an app crashes or closes during recording, the audio source disappears and system audio capture ends with it. The recording will have only what was captured before the call ended or switched context.

How to tell if the issue is permissions versus something else

If both tracks have recorded in the past but only one is recording now, the issue is likely something specific to this session — a permission that was revoked, the app losing focus, or the source changing mid-call. If this is the first attempt and one side is missing, permission is most likely. How to fix a transcript that has the speakers mixed up won't help with a completely missing voice.

The next recording after a failed one

A single failed recording doesn't affect the next one. If both permission checks pass and the app is restarted, the next recording starts fresh. The bad file stays as it is — there's no cumulative breakage from one incomplete recording to the next. Each recording is independent and will work correctly if permissions are in place.

Moving forward after checking permissions

With both permissions confirmed and the app restarted, the next recording should capture both tracks. If it still comes back with one side missing, verify both the Screen Recording permission (for system audio) and the Microphone permission (for the mic), restart the app, and try again. If both are granted and restarting doesn't fix it, the issue is more specific than permission or app state.

Comparing to other recording scenarios

If you're recording a meeting without a bot joining, both tracks should be present from start to finish. If you're recording an interview, the same two-track requirement applies. The pattern is consistent: wherever there are two separate audio sources (microphone and call audio), both need to capture.

Troubleshooting when permissions don't solve it

If permissions are all granted, the app has been restarted, and the recording still shows only one speaker, check how to record both sides of a call for additional diagnostic steps. You might also review whether the call audio was actually playing during the recording — silent calls don't produce waveforms even with correct permissions.

Why both tracks matter for the full picture

Having both tracks means the transcript captures the full conversation and context. Without system audio, transcripts become one-sided and lose the back-and-forth that makes notes useful. For how to take notes without a bot joining, both tracks are essential because they're what allow the system to understand the conversation's flow and context.

When one track failing means the whole thing fails

A complete recording requires both tracks working from start to finish. A recording where one track starts mid-call or drops out partway through creates a broken record that's harder to use than no recording at all. This is why getting both permissions right from the start matters so much.

The permission system protects everyone

The reason macOS gated system audio capture behind a permission is to protect privacy at the system level. This same mechanism protects users from apps that might try to record system audio without asking. For meeting recorder scenarios without a bot, the permission system ensures the recording stays within your control.

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. Both recording tracks work identically on both architectures. See how to record a meeting on a Mac for the complete workflow once troubleshooting is complete and both tracks are recording correctly. Understanding what a voice print is can also help you appreciate why both tracks need to record properly for speaker identification to work.

Questions

How does the app know there are two speakers if one side is missing?
It doesn't. If one track didn't record, the transcript can only see what the surviving track captured. A silent track produces no speech to transcribe, so the transcript is incomplete by design.
What causes one track to go silent while the other records normally?
Either the microphone didn't capture your side or system audio capture didn't work for the call's side. Most often it's system audio capture — the permission wasn't granted or the app restarted mid-call.
Can a transcript be reconstructed once a track is lost?
No. Audio that was never recorded can't be recovered from a transcript that was never started. The fix for future recordings is ensuring both tracks capture from the start.
Should both tracks show identical volume?
No. The microphone and system audio are different sources and naturally have different levels. Identical volume would actually be suspicious — one quiet and one loud is normal.
Does the app mix the tracks or keep them separate?
It keeps them separate by design — that's what allows speaker identification to work. One track is the microphone, the other is system audio, and both feed the transcript.
Is a missing voice the same as a mislabelled voice?
No. Mislabelled means both voices are there but one is marked with the wrong name. Missing means one track didn't record and one voice never made it into the file at all.