PERMISSIONS

How to grant the three permissions?

As a macOS menu-bar voice-input tool, VoiceInput needs Microphone, Accessibility, and Input Monitoring to work end-to-end. This is macOS's security model — every app must be authorized once.

💡 Quick reference: all three live under System Settings → Privacy & Security. On first launch, VoiceInput auto-prompts each one — just click "Allow." If you missed a prompt or a permission later breaks, follow the steps below to grant manually.

Why three permissions

VoiceInput's flow is "press hotkey → record → recognize → insert text." Each step needs one permission — none is optional:

  • Input Monitoring: listen for the global hotkey (default ++;) so VoiceInput responds in any app.
  • Microphone: capture your voice for speech recognition.
  • Accessibility: insert recognized text at the cursor position (so you don't have to ⌘V manually).

All three are local permissions — unrelated to "internet access" or "data upload." VoiceInput's free local engine runs fully offline. Even with mic permission granted, audio never leaves your machine.

1 · Microphone

For capturing your voice. Steps:

PATH

System Settings → Privacy & Security → Microphone → find "VoiceInput" → toggle on

If VoiceInput isn't listed, it hasn't requested mic access yet. Open VoiceInput and press the hotkey once to speak — macOS will prompt immediately. Click "Allow."

⚠️ Audio still missing after granting? Check System Settings → Sound → Input to confirm the right mic is selected (especially when using AirPods or external mics). VoiceInput uses the system default input device.

2 · Accessibility

Used to "insert" recognized text at the cursor (simulating keyboard input). Steps:

PATH

System Settings → Privacy & Security → Accessibility → find "VoiceInput" → toggle on

macOS may require your password or Touch ID to change this toggle.

VoiceInput not in the list?

Click the + button below the list → in Finder pick /Applications/VoiceInput.app → Add. Then toggle on.

🚨 Toggle on but still not working? A common macOS bug — turn the toggle off and back on, or remove VoiceInput from the list and re-add it. Restarting VoiceInput also helps.

3 · Input Monitoring

Introduced in macOS 13+ specifically for "listen to keyboard events." VoiceInput needs this to respond to your hotkey from any app. Steps:

PATH

System Settings → Privacy & Security → Input Monitoring → find "VoiceInput" → toggle on

💡 Of the three, Input Monitoring is the most often forgotten — its name isn't intuitive. If your hotkey fails completely, check this one first.

All three granted but still broken?

This usually means permission identity mismatch — macOS uses certificate fingerprints to identify authorized apps. If you've installed dev builds or older versions, the authorization may attach to the wrong build.

Fix

  1. Remove VoiceInput from all three permission lists (click -)
  2. Quit VoiceInput (menu bar → Quit, or +Q)
  3. Reopen /Applications/VoiceInput.app
  4. Press the hotkey once — macOS will prompt for all three permissions again. Grant in order.

Still broken? See Troubleshooting for other causes.

Do these permissions leak my privacy?

No. VoiceInput runs in the strictest privacy mode by default:

  • Free local engine is fully offline: recording, recognition, and text insertion all happen on your machine. Mic permission is only used for capture — audio never leaves.
  • History is in a local database: all recordings and AI tidy results live in ~/Library/Application Support/VoiceInput/. Delete the app and the data is gone.
  • Cloud engine is opt-in: with Pro you can switch to the cloud engine (more accurate + auto punctuation). Audio is sent to Volcano Engine for recognition and discarded — we don't store it.

Detailed privacy design will be on our blog (coming soon).