LeCoqRoti
Audio Hijack runs on any Mac with macOS 14.4 or newer, including the latest macOS 27 Golden Gate release, and works on both Intel and Apple Silicon machines. It is a routing and recording tool that treats every sound source on your system as a discrete block you can capture, process, and combine. Rather than simply grabbing system-wide output, it lets you pull audio from individual applications, connected devices, and online streams in a single session, applying real-time effects along the way. The result is a clean, deliberate recording setup that suits voice work, call archiving, music capture, and analogue preservation equally well.
- Record from Safari, Music, Skype, FaceTime, DVD players, USB microphones, and any line-in device
- Digitise cassette tapes, vinyl records, and other analogue sources through your Mac's audio input
- Run multiple sources simultaneously in one session — combine a mic with a browser stream, for instance
- Apply noise-removal filters during recording: Denoise, Declick, and Dehum clean up hiss, pops, and electrical hum
- Use the built-in Transcribe block to turn speech into text, with an updated Whisper.cpp backend for improved accuracy
- Route audio between applications without touching a physical cable
The app stays on macOS only, which means its design follows the platform's conventions closely. Version 4.6.0 refines the transcription side of things and extends compatibility forward. In short: if your work or hobby involves capturing what passes through a Mac's audio pathways, Audio Hijack gives you the routing and the controls to do it with precision.
Toolchain role: hands off recordings to an editor
Audio Hijack does not try to be a full editing suite, and that is precisely why it fits so neatly into a working day. A session set up to record a call can pass its finished file straight to the Scripting block, which hands it off to an AppleScript or shell script — open it in your editor of choice, add it to a specific folder, or attach it to a task manager entry. That small chain, from capture to handoff, is where the app earns its place: it does not ask you to reorganise your workflow around it.
Used alongside Rogue Amoeba's own SoundSource, the two apps cover different ground without overlap. SoundSource handles per-application volume and output routing for listening; Audio Hijack handles the recording side. When they run together, you can route a system alert one way and your recording target another, without either interfering. Loopback, the company's virtual-audio-cable tool, adds another layer: you can create a virtual device in Loopback that pulls from several sources, then drop that device into Audio Hijack as a single block. Podcasters who want to record a remote guest, local mic, and soundboard feed into one track often set it up just this way.
On the macOS side, the app speaks to Shortcuts through its scripting support, which means you can trigger a recording session from a menu-bar shortcut or a Focus mode change. There is also a quiet compatibility with Automator and with third-party launchers: a pre-configured session file, double-clicked, launches straight into recording. No dock icon bounce, no splash screen — the session is running before you have settled into your chair.
Resource use: 220 MB of RAM while recording ten sources
Audio Hijack 4.6.0 loads quickly on both Intel and Apple Silicon Macs, and the block-based interface assembles without any perceptible lag, even with ten or more blocks in a single session. Memory use tends to hover around 150–220 MB during a typical multi-source recording — noticeable but not heavy on machines with 8 GB of RAM or more. CPU draw depends on the processing chain: a simple application-to-FLAC capture barely registers above idle, while a session running Denoise, Declick, and the Transcribe block simultaneously will push a single core to around 25–35% on an M1 chip. That leaves plenty of room for a browser, a text editor, and a call running at the same time.
Stability across long sessions is one of the app's quiet strengths. Recordings stretching past the three-hour mark — a conference, a live stream capture, a vinyl digitisation marathon — complete without drift or dropouts. The app's handling of sample-rate mismatches deserves mention: when one source feeds 44.1 kHz and another 48 kHz into the same session, Audio Hijack resamples on the fly without clicks or pitch artefacts. That is the kind of detail you notice only when it goes wrong elsewhere.
The Transcribe block, updated in this version with the Recommended model and the newer Whisper.cpp backend, processes faster than earlier builds. A thirty-minute spoken-word recording transcribes in roughly the time it takes to make a cup of coffee — not instantaneous, but usefully quick — and the accuracy on clear speech is high enough that corrections are minor. It runs entirely on-device, so no data leaves the Mac, and it does not need an internet connection. That alone shifts it from a curiosity to a daily tool for anyone who needs searchable archives of calls or dictations.
System audio session: it captures what you hear, with a setup step
The first question is almost always about system-audio capture. People want to know: can it record whatever is playing through the speakers? The answer is yes, but the mechanism is specific. You do not simply hit a button and grab "system audio." Instead, you add an application source block — Safari, Music, a VoIP client — and target that application directly. This means you capture exactly what you intend, without notification pings or background noises bleeding in. It also means the app does not need a kernel extension or a system-audio driver that breaks with every macOS update.
Another common question concerns call recording on platforms that do not offer a built-in option. Audio Hijack sees FaceTime, Skype, and most other VoIP apps as individual sources. You can capture each side of the conversation on separate tracks — local mic in one block, remote audio in another — and balance them afterwards. The legal and ethical side is yours to manage, but the technical path is straightforward. Some people search for an audio hijack download precisely because they need a reliable call-recording setup that does not depend on a service's own cloud features.
A third point of hesitation: will it conflict with other audio tools? In practice, Audio Hijack coexists with DAWs, screen recorders, and conferencing apps without much friction. It works at the application level rather than hijacking the hardware driver, so Logic Pro or Ableton Live can still use the audio interface directly while Audio Hijack grabs a browser stream in parallel. The only occasional snag is sample-rate locking when a DAW takes exclusive control of an interface, but switching to a different input or using a virtual device usually resolves it.
Small points that are easy to overlook
There are details in the interface that reward attention. The meters in each block can be expanded with a drag, turning a thin line into a full waveform preview — useful for checking levels before you commit to a long take. The Recorder block's timer can be set to stop automatically after a given duration, which is handy for capturing a radio programme while you step away. And the block-connection dots respond to a right-click: you can mute a connection without deleting it, letting you test one branch of a complex session in isolation.
The Templates pane, often ignored after the first week, is worth revisiting. You can save a session as a template with all its filters, file-naming conventions, and post-processing scripts intact, then invoke it from the menu bar or a keyboard shortcut. A template for "interview with transcription" and another for "vinyl capture with Declick" means the setup happens once and the recording happens many times.
The built-in file-format options go deeper than the obvious MP3 and AAC choices. FLAC output supports bit depths up to 24-bit, and the AIFF and WAV settings can be configured for mono or stereo. There is even a setting to split recordings automatically by silence, which turns a long capture of an album into separate tracks without any editing. The Dehum filter, meanwhile, has frequency presets for 50 Hz and 60 Hz mains hum — a small touch that saves guesswork when digitising old tapes recorded in different countries.
Finally, the app's handling of the ACE (Audio Capture Engine) component is worth noting for anyone who has seen older versions leave behind a virtual device after uninstallation. Version 4.6.0 keeps its helper processes tidy, and the ACE no longer shows up in System Settings as a persistent audio device unless the app is actively running. It is a small piece of housekeeping, but it means the system stays clean between sessions.
Correspondence 0
Take a chair or join the table to send a letter.
No correspondence on this note yet. Letters are read before they appear.