Privacy is a core constraint. Local Whisper does not send audio or transcript text to a hosted speech API.

​
Runtime boundary

ComponentRuns at
Parakeet-TDT v3In-process MLX
Qwen3-ASRIn-process MLX
Kokoro TTSIn-process MLX
WhisperKitlocalhost:50060
Apple IntelligenceOn-device
Ollamalocalhost:11434
LM Studiolocalhost:1234 by default; private LAN allowed
History, config, backupsLocal filesystem under ~/.whisper/
Setup, model downloads, wh update, and wh doctor --fix can use the network to install packages, fetch models, or update the checkout.

​
Permissions

PermissionWhyScope
MicrophoneRecord voice for transcriptionActive during recording
AccessibilityDetect global hotkey and text shortcutsMonitors key events for configured shortcuts
The app does not request contacts, location, camera, or analytics permissions.

​
Audio lifecycle

  1. Local Whisper records a temporary WAV file under ~/.whisper/.
  2. The selected local engine transcribes the audio.
  3. Grammar text is sent only to the selected local, on-device, localhost, or private LAN backend when grammar is enabled.
  4. The result is copied to the clipboard for double-tap dictation, or pasted at the cursor for hold-to-record and auto-paste dictation.
  5. Audio is retained under ~/.whisper/ for backup according to history settings.

​
Vulnerability reporting

Report vulnerabilities privately through GitHub private vulnerability reporting: 
https://github.com/gabrimatic/local-whisper/security/advisories/new
Do not open public issues for security vulnerabilities. Include reproduction steps, demonstrated impact, and a suggested fix when available. Out of scope:
  • Issues requiring physical access to the machine
  • Issues requiring the user to grant Accessibility or Microphone permission to a malicious process
  • Prompt injection via grammar backend responses, because Local Whisper copies text and does not execute it