Skip to content

Repository files navigation

Capt

Live, on-device captions for anything your Mac is playing.

Capt taps system audio, transcribes it with Apple's on-device SpeechAnalyzer, and draws captions in a click-through overlay above every window, including fullscreen video. Capt has no servers, accounts, advertising, analytics, telemetry, crash reporting, or other tracking. Audio and caption text stay in memory and are never saved or sent anywhere.

Born from watching videos on sites that don't ship captions. Accessibility first.

Capt menu with captions inactive, English selected, and theme, text size, resize, and permission controls

Download and install

Download the latest release. The current downloads require an Apple Silicon Mac (M1 or newer) and macOS 26+.

For newer changes, PR and master preview builds are available as Actions artifacts. Only version tags publish releases; see RELEASE.md.

  1. Download the DMG from the release page.
  2. Open it and drag Capt into Applications.
  3. Open Capt, click its menu bar icon, select the spoken language, and enable Capt.
  4. Grant System Audio Recording permission when prompted. The first use of a language may download Apple's on-device speech model.

Current alpha downloads are ad-hoc signed and not notarized by Apple. If macOS blocks opening Capt and you trust this release, follow the release page's first-launch instructions or Apple's guidance.

The released app reflects its tagged source; the features documented below may include newer changes on the default branch.

Known limitations

  • Language is chosen, not detected. SpeechAnalyzer needs a locale up front. Auto-detection needs a cloud engine behind the same protocol.
  • The overlay uses the primary display. Layouts are stored per display, but Capt currently shows its caption panel only on the display macOS treats as primary.
  • Signing identity is chosen automatically. build.sh prefers a Developer ID Application or Apple Development certificate from the keychain, so the code signature (and the System Audio Recording grant keyed to it) survives rebuilds. CAPT_SIGN_IDENTITY overrides the choice; - forces ad-hoc, and the script warns that permissions may be re-requested after each rebuild.
  • Local builds are not notarized. The build script produces a hardened, signed app for local use; use Scripts/package.sh release for notarized downloads (see RELEASE.md).
  • If captions never appear and the panel says no audio is being detected, grant System Audio Recording in System Settings › Privacy & Security.

Contributing

See CONTRIBUTING.md. Security and privacy reports go to SECURITY.md.

License and legal

Capt is source-available under the Functional Source License 1.1, MIT Future License: use, modify, and redistribute it for any purpose except building a competing commercial product, and each version becomes MIT-licensed two years after release. Third-party notices are in NOTICE; Core Audio helpers are adapted from insidegui/AudioCap, BSD 2-Clause. End-user terms are in TERMS.md and the privacy policy in PRIVACY.md.

About

Generate captions live on-machine for any language, any time.

Resources

Contributing

Security policy

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages