Sound Stage

Sound Stage is a macOS prototype for arranging real app audio into a simple spatial stage.

Status: developer preview / technical MVP. The project is currently run from Swift Package Manager and is intended for prototyping, technical review, and low-level audio iteration. It is not yet packaged, signed, notarized, or distributed as a downloadable macOS app.

The app focuses on real per-process Core Audio capture:

• independent source volume
• mute and solo
• draggable 2D stage placement with preset shortcuts: Center, Left, Right, Left Front, Right Front, Left Rear, Right Rear, Far, Muted
• three scene profiles: Work, Focus, Media
• persisted source preferences
• menu bar mini panel quick controls
• per-source quality modes: Reference, Hi-Fi, Spatial, and Spatial+
• experimental Core Audio process discovery and stereo process tap capture for real app audio on macOS 14.2+

Requirements

• macOS 14.2 or later
• Xcode Command Line Tools with Swift 6 support
• System audio capture permission for process taps

Run

swift run SoundStagePrototype

Test

swift test

Prototype Boundary

The current build has one input path:

• Process Tap: an experimental Core Audio path that scans active output processes, creates a private stereo process tap for one process, and feeds that stream through a lower-coloration direct stereo mixer.

Process taps require macOS 14.2+ and system audio capture permission.

The real-audio path prioritizes fidelity over dramatic spatialization: stereo program material is preserved, reverb and obstruction coloring are bypassed by default outside the Spatial+ experiment, and the generated test tones/mock sources have been removed. Reference mode uses unity gain for a 100% center source, a 100% main mixer, and no captured-audio limiter so A/B listening is as close as this process-tap route can make it. Hi-Fi mode keeps the protective headroom, source trim, and limiting used for mixed-source listening. The process-tap buffer tracks the tap device sample rate, linearly resamples into the app render rate when needed, and reports native/resampling status plus buffered latency and underrun/overrun counts while a tap is running.

Spatial mode is opt-in per source and does not replace the Reference or Hi-Fi direct stereo route. Meeting and chat sources use a point-source HRTF route for clear placement. Browser, music, and game sources use a virtual stereo speaker pair: the captured left and right channels are read by independent mono cursors, rendered as separate HRTF point sources, and offset around the selected stage coordinate. Spatial+ is an experimental A/B profile that keeps the same process-tap path but maps placement through azimuth-first coordinates, rotates stereo pairs around the source direction, and adds a very light externalization reverb blend. The main window includes a Listening Lab control for quickly pinning active sources to the left in Reference, Spatial, or Spatial+. Source placement is draggable and head-relative for now; there is no head tracking or room-locked AirPods-style scene simulation yet.

Why This Shape

The app stays intentionally small while the low-level audio path is refined. Scan for currently audible apps, start a process tap, then place the real source on the stage.

Not Yet Included

• packaged .app build output
• code signing or notarization
• installer, DMG, or GitHub Release artifacts
• onboarding for audio capture permissions
• head tracking or room-locked spatial rendering