Skip to main content
SpeechKit for macOS is an early beta that does Dictation only. It runs on macOS 14 or newer on Apple Silicon. Intel Macs are not supported. The current public release is v0.77.19, published 2026-10-30.
The macOS bundle is signed ad-hoc, without an Apple Developer ID and without notarization. macOS quarantines it and may claim the app is damaged. It is not. Removing the quarantine label takes one documented command, described below.
In everyday terms: this is a first, limited version for Macs with an Apple chip. It turns your speech into text and types it into the app you are using. It is free to try, but macOS treats it with suspicion because it is not signed by an Apple developer account, so the install needs one extra step that you do in Terminal. Nothing below needs an administrator password.

​
Before you start

  • A Mac with an Apple chip (M-series) running macOS 14 or newer. Intel Macs are not supported.
  • A few minutes. Local speech models are downloaded later, on demand.
  • Know that Terminal is an app on your Mac (press Cmd+Space and type Terminal). You only paste one command into it.

​
Download

macOS bundle (Apple Silicon)

Download SpeechKit-macOS-arm64.zip from the newest published release.
Verify the download against SHA256SUMS.txt before opening it. That file lists the Windows assets too, so check only the line for the macOS archive:

​
Install

  1. Unzip the download and drag SpeechKit.app into your Applications folder. Move it before the next step: macOS runs an app left in Downloads from a hidden read-only copy, and SpeechKit cannot update itself from there.
  2. Remove the quarantine label in Terminal:
    Nothing is printed when it works. The command needs no password and affects only this app.
  3. Open SpeechKit from Applications and grant Microphone and Accessibility when asked.
What to expect after each step:
  • After step 1, the app sits in your Applications folder, no longer in Downloads.
  • After step 2, Terminal prints nothing. That means the command worked.
  • After step 3, SpeechKit opens without the “damaged” message, shows its item in the menu bar at the top of the screen, and macOS asks for the permissions.
Read the full unsigned-release notice for the reasoning behind each step. Nothing in the procedure asks you to disable Gatekeeper system-wide or to run anything as root.

​
What works, and what does not

The hotkey listener installs but receives no key events, so the menu-bar item is the way to start a dictation until that is fixed. Use the Windows client when you need the complete feature set.

​
Permissions after an update

macOS remembers permission grants against an app’s exact signature. Every ad-hoc signed build has a different signature, so macOS asks for Microphone and Accessibility again after each update, and sometimes for keychain access. That is expected for an unsigned beta. If a permission still seems missing, SpeechKit may also ask for Input Monitoring, which the hotkey needs. Hotkeys do not work in this build, so you can leave that one off for now. If an update leaves dictation unresponsive, open System Settings → Privacy & Security → Accessibility and switch SpeechKit back on. It picks the permission up while running; no restart is needed.

​
If something goes wrong

  • macOS says the app is damaged. It is not; the bundle is intact. Run the quarantine command from step 2 and open it again.
  • The command prints an error. Check that the app is in /Applications and that the spelling matches step 2, including the capital letters.
  • Nothing is typed into the other app. Check System Settings → Privacy & Security → Accessibility and switch SpeechKit on. Text also cannot be typed into a password field; macOS blocks it on purpose.
  • No microphone input. Check System Settings → Privacy & Security → Microphone for SpeechKit, and the input you chose in SpeechKit. See Set up hotkeys, microphone and language.
  • The hotkey does nothing. Expected in this build. Use the menu-bar item to start a dictation.
  • An update asked for everything again. Expected; see “Permissions after an update” above.
  • A speech model does not download. Check your connection and the privacy scope. See Work offline with local models and Fix common problems.
  • You want to update or remove it. See Update or uninstall SpeechKit.

​
Next steps