Mac app to record system audio
Recording what plays through your Mac's speakers is the single most counter-intuitive task on macOS - out of the box, no app can do it. Nerpa is a Mac app that captures system audio directly using Apple's modern Core Audio Process Tap: one click to record anything audible - calls, webinars, in-app players - alongside your microphone, then a speaker-aware transcript. Below: why this is hard at all, and the honest comparison of every workaround.
Why macOS makes this hard
On Windows, capturing "what you hear" has always been half-supported. On macOS, Apple's design is stricter: each app sends its audio straight to the output device, and no other process is allowed to listen in. Privacy-first for users - annoying for anyone with a legitimate reason to record a call or a webinar they're attending.
So every "system audio recorder" you've ever seen is actually one of four workarounds. Knowing which one an app uses tells you exactly what it will ask of you.
The four workarounds
- Virtual audio driver (BlackHole and friends). Re-route system output into a fake device and record from it. Free and battle-tested, but you build a Multi-Output Device by hand, and a misconfigured one can leave your speakers silent mid-call. We describe the setup honestly in the general Mac recording guide.
- ScreenCaptureKit-based apps. Apple's screen-recording API can carry audio - so some apps ask for the full Screen Recording permission just to get sound. It works, but granting an app the right to see your whole screen to record a call is a lot to ask.
- Hardware loopback. A cable from output to input, or an external interface. Bulletproof and zero software - and zero convenience.
- Core Audio Process Tap (macOS 14.2+). Apple's purpose-built API for exactly this job: tap system audio with a narrow System Audio Recording permission - no screen access, no re-routing, speakers keep working normally.
The modern way: Core Audio Process Tap
Nerpa uses the Process Tap wherever your Mac supports it (macOS 14.2+; older versions fall back to the screen-capture path automatically). What that means in practice:
- Install Nerpa and sign in once.
- Grant System Audio Recording once in System Settings - a narrower permission than Screen Recording; your screen is never involved.
- Hit record. Anything audible - Zoom, Meet, Teams, webinars, browser players, in-app audio - is captured alongside your microphone.
- Stop, and the transcription lands: speaker labels, summary, decisions, action items. The raw audio stays in a local folder on your Mac.
The same capture powers bot-free call recording - one engine for calls, webinars, and anything else that plays.
System audio, one click
Nerpa records everything your Mac plays - no drivers, no re-routing, no screen permission - and turns it into a speaker-aware transcript.
Latest stable release. Requires macOS 12+
Methods compared
| BlackHole driver | ScreenCaptureKit apps | Hardware loop | Nerpa (Process Tap) | |
|---|---|---|---|---|
| Setup | Manual Multi-Output Device | Install app | Cables | Install app, grant once |
| Permission asked | None (you re-route yourself) | Full Screen Recording | None | System Audio Recording only |
| Speakers keep working | Not always - misconfiguration silences them | Yes | Depends on wiring | Yes |
| Captures mic alongside | Manual | Depends on app | Manual | Yes, together |
| Transcript with speakers | No | Depends on app | No | Yes, automatic |
| macOS version | Any | Any | Any | 12+; Process Tap on 14.2+ |
One honest note: the Process Tap path needs macOS 14.2 or newer. On older systems Nerpa falls back to the screen-capture approach automatically - it still works, it just asks for the wider permission.
Frequently asked questions
Why can't Mac apps just record system audio?
By design: macOS sends app audio straight to the output device and lets no other process listen in. Every workaround - virtual drivers, screen-capture APIs, process taps - exists to cross that boundary, and each asks for different permissions.
What is BlackHole and do I still need it?
BlackHole is a free virtual audio driver: you create a Multi-Output Device so sound goes both to your speakers and to BlackHole, then record from it. It works but is fiddly and can silence your speakers if misconfigured. Nerpa captures system audio directly - no driver, no re-routing.
Why do some recorder apps ask for Screen Recording permission?
They use Apple's ScreenCaptureKit to tap the audio, and macOS ties that API to the full Screen Recording permission - even though no video is needed. Modern macOS (14.2+) offers the Core Audio Process Tap, which needs only a System Audio Recording permission; that's what Nerpa uses.
Can Nerpa record audio from any app?
Yes - anything audible through your speakers or headphones: Zoom, Meet, Teams, webinars, in-app audio, browser players. It's captured alongside your microphone, then transcribed with speaker labels, a summary, and action items. And if you use Obsidian, every transcript can land in your vault as Markdown - see the Obsidian integration.