Text from screen
Copy text from anything on your screen
Screen OCR for Windows and macOS: press one shortcut, drag a box over an image, a video subtitle, a scanned PDF or an error dialog, and get the text. Runs on your device, free, no upload. Then translate or rewrite it in place.
Some text refuses to be selected: subtitles burned into a video, a quote inside an image, a scanned contract, the error dialog that will not let you highlight it, a slide in a video call. You end up retyping it or taking a screenshot and squinting.
Text from screen reads it off your display. Press the shortcut, drag a box over the text, and it is yours to copy, translate, summarize or rewrite. Recognition runs on the text engine built into your operating system, so nothing leaves your computer and it costs nothing.
How to use it
- Press Ctrl+Shift+O. The screen freezes and dims, so a subtitle stays put even if the video keeps playing.
- Drag a box over the text you want. Any region, any app, any window.
- Quick Actions opens with the recognised text. Copy it, or pick Translate, Summarize or Rewrite.
- Press Add to type the result into the app you were in, or Copy to keep it on the clipboard.
Where it earns its keep
- Subtitles: freeze the frame, read the line, translate it. Works on any player, any streaming site, any video call.
- Scanned PDFs and photos of documents: pull a paragraph out without an OCR tool or an upload to a website.
- Error dialogs and installers: copy the exact message into a search or a support ticket instead of retyping it.
- Screenshots people send you: the address, the code, the table, straight into text.
- Slides in a meeting: grab the bullet points while the presenter talks.
- Then AI, if you want it: the recognised text goes into Quick Actions, so translating a subtitle is one more click.
Compared with built-in and common tools
| Built-in and common tools | KeyShift | |
|---|---|---|
| Windows | PowerToys Text Extractor, Snipping Tool text actions on Windows 11 | Any Windows 10 or 11, one shortcut |
| macOS | Live Text, inside Preview, Photos and Safari only | Any window, any app |
| Translate or rewrite the result | Copy, then paste elsewhere | Same window, one click |
| Insert into the app you are in | Paste by hand | Add types it where your cursor was |
| Where recognition runs | On device | On device, nothing uploaded |
| Cost | Free | Free, no credits for the recognition |
Tips
- Languages follow your system. On Windows, add a language pack under Settings → Time & language if a script is not recognised.
- On macOS, allow Screen Recording for KeyShift AI once (System Settings → Privacy & Security). Without it the capture sees only your wallpaper.
- For a whole page of a scanned PDF, drag column by column; the text comes out in reading order.
How KeyShift works, in any app
- Quick Actions: select text, press Ctrl+Shift+A (⌘⇧A on Mac), pick an action. The result replaces the selection.
- The typed trigger: write normally, add keyshift plus your instruction, press space twice. No selection, no shortcut.
- Voice (Ctrl+Shift+M), text from screen (Ctrl+Shift+O) and clipboard history (Ctrl+Shift+V) for the text you do not have yet.
- Undo any replacement with Ctrl+Shift+Z, up to 20 steps.
Windows 10 and later, macOS 11 and later, and Android. See the getting started guide.
Questions people ask
Is the screenshot uploaded anywhere?
No. Recognition uses the OCR engine built into Windows or macOS, on your machine. Only the action you run afterwards, such as a translation, sends the recognised text to the AI.
Does it cost credits?
Reading the text is free and uses no credits. Translating or rewriting it afterwards is a normal AI action.
Can it read handwriting or Chinese?
Printed text in any language your system has a language pack for. Handwriting depends on the OS engine and is hit and miss.
Does it work with two monitors?
Yes. The freeze covers every display, and you drag on whichever one holds the text.
Try it in your own apps
Free plan, 25 AI actions a month, no card needed. Pro from €4 a month billed yearly.