FAQ
Does my audio leave my Mac?
No. Dictation audio and the transcript stay on your Mac. The speech model is downloaded once, and after it is installed dictation can work offline. Purchase, model download and update checks still use the internet when needed.
More detail in Privacy on Device.
Does it work in any app?
Phantom Voice inserts the finished transcript into the focused text field system-wide, so it is designed for editors, terminals, issue trackers, documents, chat and more.
See Dictate Into Any App.
What happens while it is idle?
The native menu-bar agent stays out of the way with near-zero CPU use while idle. Its selected whisper.cpp context remains resident after warm-up so the next dictation can start quickly. Memory use depends on the model you choose.
What Mac do I need?
macOS 14 or later on Apple silicon, and it transcribes English. Large v3 Turbo is the recommended default, with Base, Small and Medium available when you prefer a lighter model.
On first run it asks for three macOS permissions: Microphone to hear you, and Accessibility plus Input Monitoring so the push-to-talk hotkey works system-wide. Keep Phantom Voice in your Applications folder, or macOS will not remember them.
What is optional local cleanup?
Cleanup can refine punctuation and apply your personal dictionary using an additional local model of about 1.1 GB. It is off by default, and it never sends the text to a hosted AI service.
See Optional Cleanup.
How do licenses and refunds work?
Voice and each paid indicator use their own Phantom ID. There is no account or subscription. Phantom Voice includes a 14-day trial; the $3 indicator styles do not have one, since you can watch every one of them run at full size before you buy. Purchases are covered by the refund policy.
See Pricing and Licenses.
Next#
- Install Phantom Voice — get set up
- Troubleshooting — for problems after install
- Back to the documentation home