Enter screenshot mode explicitly

In Actions → Screenshot OCR → gear, configure a modifier-plus-key combination. Use that shortcut, the menu-bar Screenshot OCR command, or the widget action, then drag a region containing text. The old hold-key gesture is removed; configure the new shortcut when upgrading.

OCR does not keep a global mouse interceptor active while waiting for you to start. Normal dragging in other apps should not enter screenshot mode.

Choose an OCR model

Choose On-device — Recommended for local recognition or Microsoft — Online for online recognition. Small type, low contrast and complex layouts can still produce errors; review the result. If Microsoft cannot process a very small region, capture more space around the text or choose on-device recognition.

Review, edit, or copy

Choose a result popup or Copy to Clipboard in the action settings. In the popup, correct recognized text before copying or playing it. Optional playback uses local macOS speech voices. On-device recognition needs no online service; Microsoft OCR is an optional online choice.

Leave screenshot mode

Press Escape or right-click to cancel. Disabling Screenshot OCR on the Actions page, or pausing ActionClip, cancels capture and disables OCR listening immediately; no restart is required.

If capture is blocked

Use Open Settings in the error card and allow ActionClip under macOS Privacy & Security → Screen Recording (wording varies by macOS). macOS may require quitting and reopening ActionClip. Accessibility is also needed for ActionClip's selected-text workflows. If recognition is empty, retry with a larger, clearer region.

Translate a screenshot automatically

In ActionClip 2.3.2 or later, open Actions → Screenshot OCR → gear. Under Translation, turn on Auto translate and choose Translate to. OCR has its own target-language choice and uses the provider configured for the Translate action.

Recognition remains available with Auto translate on. Choose your OCR model independently of translation and result destination. Under Result, choose Show Result Popup to review the translation or Copy to Clipboard to copy it directly. Turning off Auto translate returns the recognized text without translation.

On-device recognition supports the languages available through macOS and happens on your Mac. With an online translation provider, recognized text is sent to that provider; screenshot pixels are not sent for on-device recognition. Microsoft OCR uploads the captured image for recognition. Online translation requires Allow online actions. Local translation depends on the languages available on your Mac.

Clean up line breaks and restore the original

Turn on Cut line breaks in the Screenshot OCR action’s gear settings to join wrapped lines while retaining paragraph breaks. The result popup offers Restore original text so you can return to the recognized layout after cleanup.

Cleanup is independent of recognition accuracy. Balanced returns ordinary text quickly and briefly waits for Live Text when lists or tables are detected. More Accurate waits longer for Live Text to better preserve bullets, numbering, punctuation and tables; it can take a few seconds longer. Both recognize the screenshot locally. Optional online translation sends recognized text, not screenshot pixels, to your configured provider.

Choose on-device or Microsoft recognition

In ActionClip 2.3.6, open Actions → Screenshot OCR → gear → OCR model. On-device — Recommended recognizes the captured region on your Mac. Microsoft — Online sends the image of that region through ActionClip’s service to Microsoft for recognition.

Microsoft recognition requires Allow online actions; when online actions are off, ActionClip uses Apple OCR on this Mac. On-device recognition improves handling of small text, but both providers can make mistakes. Capture a clear region and review names, numbers and formatting.

Recognition and translation are separate choices. Enable Auto translate to translate the recognized text with your translation provider. The result popup shows the original recognition and translated text together. Online translation sends the recognized text to that provider, independently of your OCR model.

Back to all guides