Why can’t I search or copy text in my PDF?

Because the file is a scan: a photograph of a page. Pictures contain no characters, so there is nothing for search or copy to find. Text recognition (OCR) fixes it by writing an invisible, selectable text layer over the image.

Will OCR change how my document looks?

No. The page pixels are untouched; the recognized text sits in an invisible layer underneath your picture. The file simply gains working search, select and copy.

On printed text, high — our printed test page reads at confidence 93/100, and your own score is reported after every run. On handwriting, honestly poor: 41/100 on our real handwritten test, and the tool tells you rather than pretending. Skim the result against the image before relying on it, especially numbers.

Which languages can it read?

English (2.0 MB model) and Simplified Chinese (1.7 MB model) at launch — one model loads at a time, from this site, never from a third-party CDN.

Is my scan uploaded to be recognized?

No. The OCR engine runs in your browser as WebAssembly; the scan, the recognition and the output never leave your device.

You can’t search your PDF because it’s a picture

Search finds nothing, the cursor won’t select, copy comes back empty — because a scanned PDF is a photograph of words, and a photograph has no text to search. The fix is OCR: read the picture, lay an invisible text layer over it. It happens below, on your device; the scan is never uploaded.

The costs, first: the OCR engine is 1.5 MB and your language model loads with it — English 2.0 MB, Simplified Chinese 1.7 MB. Measured, not estimated. Both load from this site, only when you run it.

This tool does its work with JavaScript, entirely inside your browser — that is the design: no upload, no server, your file never leaves your device. With JavaScript off, the tool can’t run, but every claim on this page still can be checked: the source is public, and the page itself makes no third-party requests of any kind.

Drop a PDF here — it stays on this device. The controls you get come from the file, not from a menu.

All the tools

Why search can’t see your document

A PDF page can carry text two ways: as text — characters a viewer can search, select and copy — or as pixels, a picture of a page. A scanner or a phone camera produces the second kind. To search, your viewer needs characters, and a scan hasn’t got any; it has a photograph of some. That is the whole problem, and no setting in your viewer changes it.

What OCR does about it

Text recognition reads the picture and writes an invisible, selectable text layer over it. The page looks identical afterwards — the pixels are untouched — but search, select and copy now work, because the characters finally exist in the file. On our printed test page the recognizer reads near-perfectly (confidence 93/100, reported to you after the run).

The honest limits, before you run it

The recognizer is built for printed text; on a real handwritten page it scored 41/100 and the tool said so rather than hand over nonsense as a result. And a recognizer can be confidently wrong — an 8 read as a B can score well — so skim the recognized text against the image before relying on it, especially for numbers. The confidence score is put in front of you precisely so you know how much skimming is warranted.

If your PDF already has real text

If you can select text in some places but want it out of the file, that is not a scan problem — extract the text directly, no recognition needed. That page also tells you honestly, per page, whether it found text or a photograph.

Questions people actually ask

Why can’t I search or copy text in my PDF?: Because the file is a scan: a photograph of a page. Pictures contain no characters, so there is nothing for search or copy to find. Text recognition (OCR) fixes it by writing an invisible, selectable text layer over the image.
Will OCR change how my document looks?: No. The page pixels are untouched; the recognized text sits in an invisible layer underneath your picture. The file simply gains working search, select and copy.
How accurate is it?: On printed text, high — our printed test page reads at confidence 93/100, and your own score is reported after every run. On handwriting, honestly poor: 41/100 on our real handwritten test, and the tool tells you rather than pretending. Skim the result against the image before relying on it, especially numbers.
Which languages can it read?: English (2.0 MB model) and Simplified Chinese (1.7 MB model) at launch — one model loads at a time, from this site, never from a third-party CDN.
Is my scan uploaded to be recognized?: No. The OCR engine runs in your browser as WebAssembly; the scan, the recognition and the output never leave your device.

We invite you to double-check our work

Open your browser’s developer tools, switch to the Network tab, and run any tool on this page and nothing leaves your device.

The whole source is on GitHub. Read it.

Found a bug, or a claim that doesn’t hold up? Tell us: pdf@geodeforge.com. An example file turns a guess into a quick fix — and if yours is too sensitive to share, we understand; a description still helps.

That’s the difference between a privacy policy and a fact.

More from GeodeForge

GeodeForge is a one-person studio that crafts open-source tools that do what they say, and let you check — in your browser.

GeodeQR — QR codes that never expire GeodeCode — clean code and text before you paste it GeodeQuill — teach your AI app to write in your voice

Check at geodeforge.com →