What is macOS ScreenCaptureKit Audio?

ScreenCaptureKit is Apple’s modern, high-performance operating system framework for capturing display contents and audio streams on macOS. Introduced in macOS 12.3 Monterey and substantially expanded in macOS 14 Sonoma, ScreenCaptureKit replaced older, deprecated capture mechanisms (like AVCaptureScreenInput and CGDisplayStream).

For years, the greatest engineering hurdle for desktop note applications on macOS was recording meeting conversations: How could an app capture incoming speech from Zoom, Google Meet, or Slack without forcing users to install complex virtual sound card drivers or inject intrusive bot accounts? ScreenCaptureKit solved this problem permanently.

Architectural mechanics of ScreenCaptureKit

ScreenCaptureKit operates directly within the macOS WindowServer and CoreAudio infrastructure, delivering high efficiency and security:

[Zoom / Meet / Teams Call Audio]
               ↓
    [macOS CoreAudio Subsystem]
               ↓
      [ScreenCaptureKit SCStream]
               ↓
    [SCStreamOutput Audio Delegate]
               ↓
[PCM Audio Buffers → Local Whisper ASR]

1. Granular capture filters (SCContentFilter)

ScreenCaptureKit allows applications to capture specific target applications (e.g., capturing only Zoom’s audio while ignoring Spotify music playback in the background), or capture all system audio streams simultaneously.

2. High-performance stream configuration (SCStreamConfiguration)

Applications configure audio capture parameters with a few lines of code:

let config = SCStreamConfiguration()
config.capturesAudio = true
config.sampleRate = 48000
config.channelCount = 2
config.excludesCurrentProcessAudio = true

3. Explicit user consent and security

macOS protects user privacy by requiring explicit user authorization in System Settings > Privacy & Security > Screen & System Audio Recording. The operating system displays a visible menu bar recording indicator whenever a ScreenCaptureKit stream is active.

4. Zero-copy audio buffer lifecycle (CMSampleBuffer)

When ScreenCaptureKit delivers audio packets, it packages them inside CoreMedia CMSampleBuffer instances. Highly optimized applications convert these incoming buffers into raw AVAudioPCMBuffer floats without making unneeded heap memory copies, ensuring uninterrupted audio streams even during system load spikes.

ScreenCaptureKit vs legacy macOS audio capture techniques

To appreciate ScreenCaptureKit, consider how audio was captured on macOS previously:

Capture Method	Driver Type	Stability & Kernel Risk	Setup Friction	Latency
ScreenCaptureKit	Native User-Space API	Zero (No kernel extensions)	Single OS permission prompt	< 5 ms
HAL Audio Plug-in	CoreAudio HAL Driver	Low (Can crash system CoreAudio daemon)	Requires installer script	~15 ms
Virtual Sound Cards (BlackHole)	Virtual Audio Loopback	Medium (Requires Aggregate Device setup)	High manual configuration	10–25 ms
Cloud Bot Injection	Webhook / SIP Client	Zero on host (Severe privacy risk)	Requires admin meeting invite	Network dependent

ScreenCaptureKit obsoleted virtual audio drivers by integrating audio taps directly into the macOS graphics and audio compositing engine.

How Oyma uses ScreenCaptureKit

Oyma leverages ScreenCaptureKit to provide completely botless meeting recording:

• When a video conference starts in Zoom, Google Meet, Microsoft Teams, Webex, or Slack, Oyma captures the incoming call audio stream via ScreenCaptureKit.
• Concurrently, it captures your own voice via the microphone input through standard CoreAudio interfaces.
• Both streams are combined in memory and fed into on-device Whisper models for immediate transcription.
• The resulting text is written directly into your Markdown vault alongside structured timestamps, action items, and speaker quotes.

There are no virtual audio cables, no third-party bots, and no cloud uploads. Everything executes securely on your Mac.