Snipping Tool OCR: Copying Text From a Screenshot
By Schema LawPublished Updated
The Windows Snipping Tool has OCR built in. After you capture a snip, the Text actions button reads the words in the picture so you can select and copy them, copy all of them at once, or black out email addresses and phone numbers. The text goes to your clipboard. Microsoft's page describes no file with a text layer.
Everything here about the Snipping Tool comes from Microsoft's support page on how to use Snipping Tool to capture screenshots. That page gives the text feature a single paragraph, so where it is silent this post says so and tells you what to try on your own machine.
Does the Windows Snipping Tool have OCR?
Yes. Microsoft's Snipping Tool page says that once you have captured a snip, you "select the Text actions button to activate the Optical Character Recognition (OCR) feature", and that this "allows you to extract text directly from your image". That is the vendor's own name for it: OCR in the ordinary sense, reading letters out of a picture.
Two limits on that answer are worth knowing before you go looking for the button.
- It works on a snip, meaning a picture of your screen. The tool reads whatever you captured: a page of a PDF open in a viewer, a frame of a video deposition, an error dialog, a table on a web page. It does not open a document and read the document.
- Microsoft's page documents it under Windows 11. The support page has a Windows 11 part and a Windows 10 part. The paragraph on copying and redacting text sits in the Windows 11 part, and the Windows 10 part lists no Text actions button. Microsoft does not state a minimum version for the feature on that page. If the button is missing on your computer, check which Windows release and which Snipping Tool version your firm has installed.
Where the Text actions button appears after you take a snip
The Text actions button appears in the Snipping Tool window, alongside the other tools for marking up an image. Microsoft's list of what you can do after capturing an image snip includes the pen, highlighter, shapes, eraser and crop, and then "Select Text actions button to extract and copy desired text from the snip."
Getting to that window takes one capture. Microsoft documents two ways to start:
- Open the Start menu, type
snipping tool, and open the app. Choose a snipping mode (Rectangle, Window, Full screen or Freeform) and select New. - Or press the Windows logo key + Shift + S to open the capture overlay directly.
Microsoft says a captured snip is "automatically copied to the Snipping Tool window where you make changes, save, and share". The text feature works from that window. The page does not say there is anything to switch on in Settings first, and it does not describe the button's exact position. Look along the toolbar of the window that holds your snip.
One practical point follows from the fact that the tool reads a picture of your screen. What you capture is what gets read. Before you snip a page of a scanned PDF, zoom the viewer so the passage fills the window and the letters are large and sharp. The post on image files as OCR input explains why a screenshot of a small, zoomed-out page reads worse than the page itself.
Copy all text and Quick redact in the Snipping Tool
Copy all text puts every word the tool recognized in the snip on the clipboard. Quick redact covers email addresses and phone numbers. Microsoft describes three choices once Text actions is on: "select and copy specific text", or use the tools "to Copy all text or to Quick redact any email addresses or phone numbers in the snip."
| Command | What Microsoft says it does | What to check yourself |
|---|---|---|
| Select and copy | You pick specific text in the snip and copy it | That the selection begins and ends where you meant it to |
| Copy all text | Copies all the text in the snip | The order of the pasted lines when the snip held columns or a table |
| Quick redact | Redacts "any email addresses or phone numbers in the snip" | Whether every address and number was caught, and what else on the screen is private |
Read the scope of Quick redact exactly as written. Microsoft names two kinds of data, email addresses and phone numbers. It does not mention names, account numbers, Social Security numbers, dates of birth or street addresses, so do not expect the command to find those. Treat it as a convenience for a screenshot you are about to paste into a chat or an internal email. A redaction that goes into a court filing or a production is a separate job with its own rules, done in a tool built for that and checked by a person.
A short worked example. A paralegal at a fictional firm, Okafor and Lind, has a scanned lease open in a PDF viewer and needs the notice address clause for an email to the supervising attorney. She zooms to the clause, presses Windows logo key + Shift + S, drags a rectangle around the paragraph, selects Text actions in the Snipping Tool window, chooses Copy all text, and pastes into the draft. She then reads the pasted paragraph against the scan, because a suite number or a ZIP code with one wrong digit sends a notice to the wrong place.
After you paste, check the clipboard text for the notice address, the party names on the lease or other contract, and any invoice figure inside the selection. If the snip includes a pleading caption, compare the case number with the screen before you close the tool. Write the document name and page beside the paste, since the text itself carries neither.
Where does the Snipping Tool process the text it recognizes?
On your computer, according to Microsoft. The support page ends its paragraph on the text feature with this sentence: "All text recognition processes are performed locally on your device." That statement covers text recognition in the Snipping Tool. It is the whole of what the page says on the subject.
For a lawyer, that sentence answers one question and leaves others open.
- What it answers. The recognition step that Text actions performs is described as local.
- What it does not cover. The same page describes other things a snip can do: it is saved automatically to your Screenshots folder, it can be shared from the See More menu, and a video snip can be sent to Clipchamp for captions. Microsoft's sentence about local processing is about text recognition, so do not stretch it over those features. If your Screenshots folder sits inside a synced cloud folder, the picture of the client's page is going wherever that folder syncs.
A snip of a privileged document is itself a copy of part of that document. Clear out the Screenshots folder as part of closing the task, or change the automatic save setting, which Microsoft says is under See More and then Settings.
Why a snip of a scanned PDF page gives you clipboard text and not a searchable file
The Snipping Tool hands you words on the clipboard, and the scanned PDF you photographed is unchanged. Clipboard text is a loose copy of some characters. A searchable file is the document itself with recognized text stored inside it, lined up behind each page image. Microsoft's page describes extracting and copying text and nothing about writing text into a file.
The difference shows up as soon as you need more than a quotation.
| You need | Clipboard text from a snip | Text layer written into the file |
|---|---|---|
| One paragraph for an email or a draft | Enough | More than you need |
| To press Ctrl+F in the PDF tomorrow | No. The PDF is still an image | Yes |
| The same result for every page | One capture per screenful, pasted by hand | The whole file in one pass |
| Page numbers that travel with the words | Lost. The clipboard holds no page reference | Kept. Each word sits on its page |
| Someone else to search the document | No. The text lives in your paste, not in their copy | Yes, in any copy of the file |
Three consequences matter in a matter file. First, clipboard text has no link back to its source, so note the document and page yourself at the moment you paste. Second, the tool reads a screen capture, so the result depends on your zoom level and monitor, and the same page can copy differently on two machines. Third, a 60-page scan would take dozens of captures, each pasted and each proofread, and you would still have an image-only PDF at the end.
So use the Snipping Tool for a passage, and convert the document when the document is the point. The guide to making a scanned PDF searchable covers how a text layer is added to a whole file and how to confirm it is there. In Schema Law you upload the PDF, the pages are read into text and tables, and searchable PDF is one of the formats listed under export formats. Read names, dates and amounts against the original whichever route you take.
If you have been using the PowerToys utility for the same job, Microsoft's own documentation now points you to the Snipping Tool. The post on PowerToys Text Extractor explains what that utility does and what Microsoft's note says.