Record system audio on a Mac.

The problem: QuickTime hears your mic, not your Mac

QuickTime Player’s audio recording captures a microphone, not what your Mac plays. For years the standard workaround was a virtual audio driver such as BlackHole, Loopback, or the late Soundflower. You install a fake output device, route the system through it, aggregate it with your mic, and remember to undo it all so your speakers work again. It works, but the setup is fragile and reroutes your actual audio path just to observe it.

The native way: Core Audio process taps

macOS 14.4 introduced a first-party API for this: with the user’s permission, an application can tap audio output without kernel extensions, virtual devices, or rerouting. echo99 is built on that API. Your sound keeps playing through whatever device you chose while echo99 taps the stream on the side and writes it to disk. You grant two permissions, microphone and system audio recording, and change nothing in your audio setup; getting started walks through both and your first recording. The API does not exist before macOS 14.4, which is why echo99 requires macOS 14.4 or later.

What you get

• system.wav: everything your Mac plays: calls, webinars, browser audio, any app.
• mic.wav: your microphone, captured at the same time but kept separate.
• Two tracks, not a mixdown. The tracks are never mixed, so you can process, edit, or archive each side independently, and transcription can label your voice “Me” and everyone else “Them-1”, “Them-2”, …
• Device-switch resilience. Unplug headphones or swap to AirPods mid-recording and echo99 rebuilds its capture around the new device, converting formats on the fly and appending to the same file, so each track stays one continuous stream.
• On-device transcript of both tracks after you stop, with speaker labels and timestamps. Recordings you already have, such as an MP4 or M4A, can be imported and transcribed the same way.

Limits

• The current version taps all system output, not a single chosen app. If music plays during your call, it lands in system.wav too, so pause it before you start.
• There is no system-only mode: echo99 always records both tracks. Muting yourself in a call app doesn’t silence mic.wav, because echo99 reads the microphone directly.
• WAV is uncompressed PCM (24 kHz, mono, 16-bit): about 2.9 MB per minute per track, so a one-hour call takes roughly 350 MB for both tracks. Clean for editing and archiving, but budget disk space for very long sessions.
• Audio only: echo99 does not record your screen.

Tell people you are recording

echo99 never joins the call, so nothing like a call app's recording banner tells the others you are recording. Many jurisdictions require the other side’s consent, and your organization or the host may have rules of its own. Say it out loud at the start, before the conversation gets going.

Related guides: Record Zoom calls, Record a webinar, Record FaceTime calls, Speaker identification, Meeting summaries, Transcribe audio and video files