Start, stop, cancel
One button starts recording, using whichever preset is currently active — there's no setup to do mid-conversation. Once it starts, the button turns into a red Stop, and a Cancel button appears next to it. The elapsed time counts up, and a live waveform shows amplitude for each enabled audio source: a trace for the microphone while recording, a separate trace for system audio if that's also on. A small dot next to the waveform pulses when speech is currently detected and settles to grey when it's quiet, so you can tell at a glance that the app is actually picking up your voice.
Stop finalizes the recording and hands it off to on-device transcription and diarization. Cancel deletes it outright, after a confirmation — there's no undo once you confirm.
While a recording is running, it's pinned at the top of the Home screen so it's always one click away, wherever else you are in the app.
Choosing your audio devices
Settings → Audio Devices is where you pick which microphone Cohearo records from, and, on desktop, which system-audio device to capture. Each has a refresh button to rescan for newly connected devices, and a test control so you can confirm a device is live before you commit to it — no need to start and discard a real recording just to check.
Changing either device marks the tab dirty and shows a save bar; both dropdowns and their tests are disabled while a recording is active, so a device can't be swapped out from under an in-progress capture.
System audio capture
Cohearo can capture what your machine is playing — a call, a video, another app — alongside your microphone, so both sides of a conversation end up in one transcript. System audio capture is available on desktop (Windows, macOS, Linux); on mobile there is no system-audio toggle, since only the microphone is ever available there.
A preset needs at least one audio source enabled on desktop; turning off microphone and system audio at the same time isn't allowed — turning one off automatically turns the other back on.
Works offline
Recording, transcription and diarization all run on your device, not on a server, so none of it needs a network connection to work. That holds in both On-device mode (nothing ever leaves the machine) and Cloud mode (the same on-device processing, with only the finished transcript syncing afterward) — the network is only needed for that sync step, not for the recording itself. See On-device vs Cloud for the full difference between the two.
How long audio sticks around
Recorded audio is kept on your device for 72 hours before it's automatically cleaned up. Within that window you can play it back or export the original file; after it expires, the transcript itself is unaffected, but there's no audio left to export or replay. This 72-hour window is the same on every plan — it's a local privacy default, not something billing changes.
On this page