Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.hyprcore.ai/llms.txt

Use this file to discover all available pages before exploring further.

The Audio panel controls every audio device Hyprcore touches—your input mic for dictation and meetings, your output device for feedback sounds, and the lifecycle of how the mic gets opened.

​
Microphone

  • Input device. Pick which microphone Hyprcore uses. Defaults to your macOS default input. Hyprcore re-scans devices when you open this dropdown so AirPods, USB mics, and Bluetooth headsets show up immediately.
  • Always-on microphone. Keep the mic stream open between dictations to reduce the latency of the first word. Useful on a quiet machine; on a Mac that’s tight on resources, leave it off.
  • Mute system audio while recording. When dictating, Hyprcore can mute your speakers so you don’t get distracted by notifications mid-thought. On by default.

​
Output

  • Feedback sounds. Hyprcore plays a short tone when recording starts and stops. Toggle the start sound, stop sound, or both.
  • Custom sounds. Drop your own .wav or .mp3 files in ~/Library/Application Support/ai.hyprcore.desktop/sounds/. Pick them in the dropdown.

​
Voice activity detection (VAD)

  • VAD enabled. Silero VAD trims silence from your audio before transcription. On by default. Turn off only if you have a specific reason—the model handles silence less well than the VAD.
  • VAD sensitivity. How aggressive VAD is at trimming. Default is balanced. More aggressive cuts more silence but can clip the start of words. Less aggressive keeps every breath.
  • Hysteresis. A small window of audio kept around speech edges so the model doesn’t lose context. Don’t change unless you understand it.

​
Levels and metering

The recording overlay shows a live waveform. If you don’t see it move when you talk:
  1. Check the right input device is selected.
  2. Check macOS hasn’t muted Hyprcore in System Settings → Sound → Input.
  3. Check Microphone permission is granted to Hyprcore.

​
Multi-channel and high-sample-rate mics

Hyprcore captures at 16 kHz mono internally—the rate Whisper and Parakeet expect. If your mic captures at 48 kHz or 96 kHz, audio is resampled (rubato) before transcription. Multi-channel mics (USB interfaces with multiple inputs) are downmixed to mono. There’s no quality loss for typical voice content.

​
System audio (for meetings)

System audio capture is configured under Settings → Meetings → Capture mode, not here. Audio settings on this page only affect the microphone.