Audio Hijack 4.6.0 — Mac version

Audio Hijack 4.6.0 landed late August 2026, bringing baked-in support for the freshly minted macOS 27 Golden Gate. The built-in Transcribe block also got a polish, bumping its Whisper.cpp backend to 1.9.2 and adding a Recommended transcription model that's a sensible starting point for voice recordings.

macOS 14.4 to 27, Intel and Apple Silicon

• macOS 14.4 through to macOS 27
• Runs natively on Intel and Apple Silicon Macs
• macOS only — no Windows or Linux build

Grab audio from any app, mic, or turntable

At its core, the app grabs audio from practically any source on your Mac. That means Safari streams, iTunes, VoIP calls, DVD players, a USB microphone, or that dusty turntable plugged into an audio interface. You can pull from several sources at once in a single session. Templates get you started quick for common tasks, and the built-in processing blocks let you tidy things up after capture. If you're digitising a box of old tapes or vinyl, the analogue-to-digital workflow is straightforward and fuss-free.

Change the default save folder before recording

Out of the box, Audio Hijack saves recordings as MP3 files to your Music folder. That works, but a few quick tweaks make life easier later.

Head to the app's preferences and change your default save location to a dedicated folder—something like ~/Audio Recordings keeps things tidy. While you're there, switch the default file format to FLAC if you plan on editing later. FLAC gives you a lossless master to work from, and you can always squash it down to MP3 when you're done.

The Global Recording Timer is easy to overlook. Set a hard stop for recordings that might otherwise run away—handy if you leave a stream capturing overnight. You'll find it under the General tab. Tick "Stop recording after" and pick a sensible ceiling. An hour does for most calls; three hours covers a long podcast session.

One more worth flipping: in the Advanced preferences, enable "Keep recording if source drops". Without it, a momentary blip in a VoIP call or a brief network wobble kills your whole session. With it, you get a gap instead of a dead file.

Start a session from the menu bar

• Quick-record from the menu bar. Right-click the Audio Hijack icon and pick any session you've saved. No need to bring the full window up. You can start, pause, and stop without ever leaving what you're doing.
• Session templates as starting blocks. The template picker gets you close, but dragging extra blocks onto the canvas is where the real speed lives. Need to record a Skype call and run Dehum at the same time? Grab the VoIP template, drop a Dehum block onto the chain, and you're rolling in ten seconds.
• Per-block level meters. Each block has a tiny meter tucked into its corner. Click it to watch the signal as it passes through. Brilliant for spotting where a filter is clipping before you waste a whole recording.
• Cmd-Shift-R starts your last-used session instantly. No clicking around. Muscle memory pays off fast.

The Transcribe block's new Recommended model is worth a look too. It's not perfect, but for clean speech it turns a recorded interview into searchable text without sending anything to the cloud. The bump to Whisper.cpp 1.9.2 means fewer garbled sentences than earlier builds.

Workflow: recording a FaceTime call for a podcast

Say you're interviewing someone remotely for a show. You want both sides of the conversation captured clean, with a bit of noise cleanup and a transcription for show notes later. Here's one way to do it, soup to nuts.

1. Set up the session. Open Audio Hijack and pick the VoIP Call template. It drops in an Application source block pre-set to FaceTime and a Recorder block pointed at your default output folder. Add a Denoise block between them—drag it straight onto the connecting line.
2. Check your levels. Make a test FaceTime call to a willing friend or just to your own voicemail. Watch the meters on each block. The Denoise block can be a little aggressive at its default setting; click it and dial the reduction back to about 30% if it's thinning out voices.
3. Add transcription. Drag a Transcribe block onto the canvas and connect it to a second output from the Application source. That way the recording and transcription run side by side. Set the model to Recommended and pick your language. The text file lands alongside the audio when you stop.
4. Name your file sensibly. In the Recorder block, click the file name and tag it with the guest name and date—Guest-Name-2026-09-15. Future you will be grateful.
5. Hit record, do the call. The menu bar icon shows a red dot while you're capturing. A quick glance tells you it's still running. Chat naturally; the Denoise block handles fridge hum and keyboard clatter in the background without much fuss.
6. Stop and tidy up. Press stop, grab the FLAC from your recordings folder, and open it in your editor of choice. The transcription file is a plain text sitting right next to it—rough but good enough for pulling quotes and timestamps.

If FaceTime throws a wobbly mid-call, that "Keep recording if source drops" setting from earlier means you won't lose everything. The recording just picks up when the app comes back.

• Added support for macOS 27 (Golden Gate)
• Transcribe block now includes a Recommended transcription model
• Whisper.cpp backend updated from 1.7.4 to 1.9.2

• macOS 14.4 or newer
• Intel or Apple Silicon Mac