How to make a scanned PDF or a photo of a document searchable
To make a scanned PDF searchable, open Make PDF searchable, select Choose a PDF or photos, pick the Language of the text and select Make searchable →. The tool reads the words on each page with text recognition (OCR) and lays them, invisible, over the picture of the page. The page looks the same, but you can now search it, select words and copy them. Photos of documents work the same way, and nothing is uploaded.
A scanner or a phone camera saves each page as a picture. A PDF reader can’t search a picture, so Ctrl+F finds nothing and copying gives you nothing. OCR fixes that.
Step by step
- Open Make PDF searchable on your phone or computer.
- Select Choose a PDF or photos. Pick one PDF, or up to 25 photos (JPG, PNG, WebP or iPhone HEIC). On a computer you can drag the files onto the page.
- Check the Language of the text. It starts on the language of the page you are on, so switch it if the document is in another language.
- For a PDF, leave Skip pages that already have text ticked. Pages that already have real text are left alone, and only the scanned ones are read.
- For photos, use the arrows to put the pages in order. Each photo becomes one page.
- Select Make searchable →. The bar shows which page is being read, and Cancel stops it.
- Select Download result ↓ for the searchable PDF. For the text alone, use Download text (.txt) ↓, Download Word (.docx) ↓ or Copy text.
The limit is 50 pages or 25 photos at a time, and 30 MB. For a longer document, split it first with Split PDF and run each part.
How accurate it is: our test
On November 4, 2026, we ran 35 real scanned pages through the tool in Chrome on a computer: five pages for each of the seven languages it reads. The pages are public-domain books and documents from Wikisource, where volunteers have typed and proofread every page, so the result could be compared with a checked text. One Hindi page was a phone photo with a shadow across it.
| Language | Characters right | Words right |
|---|---|---|
| English | 98% | 97% |
| Spanish | 96% | 93% |
| Portuguese | 97% | 92% |
| French | 95% | 93% |
| Hindi | 91% | 87% |
| Indonesian | 96% | 91% |
| Swahili | 97% | 89% |
These are the figures for the invisible text layer in the PDF. The Word copy came out about the same. Each page took about 2 to 3 seconds on the test computer.
The scores are a little lower than what you will see in practice, because the reference text leaves out page numbers and running headers, which the tool reads anyway, and a word split by a hyphen at the end of a line counts as wrong. The Indonesian pages use spelling from the 1920s, and the Swahili pages come from a book printed in 1870, so a modern letter or form should read at least as well.
In short: on a clear printed page, almost every word is right, but not every word. Names, numbers and dates are where a single wrong character matters most, so check those before you rely on the text.
Photos of paper
A photo of a letter or a form works, but a straight, well-lit photo reads much better than one taken at an angle. Before reading, the tool makes a cleaned-up grayscale copy of each page to help recognition. The image you get back in the PDF is your original photo, unchanged.
If the photo shows a table or a desk around the paper, use Scan to PDF instead. Select Choose photos, keep Find page edges automatically ticked, tick Make text searchable (OCR) and select Create PDF →. It crops and straightens the paper first, then adds the text layer in the same step.
Making a searchable scan smaller
Scans are often large. Compress PDF can shrink them and keep them searchable. Set Compression mode to Image-based — for scans and photos or Target size — for upload limits, tick Keep text searchable (OCR), then select Compress PDF →. These modes turn each page into a new picture, which would normally drop the text, so the tool reads the text again and adds it back.
In our test, a five-page real English scan of 692.9 KB with a 300 KB target came out at 235.8 KB, and 98% of the characters in the text layer were right.
What the text download is good for
- The searchable PDF is the one to keep or send. It looks like the original, and anyone can search and copy from it.
- The .txt file and Copy text give plain text, for pasting into an email or a form.
- The Word copy has the text of each page as paragraphs, with a page break between pages. It doesn’t rebuild the layout, tables or pictures, so expect to tidy it up.
Privacy and offline use
The pages are read on your own device. The first time you use a language, the recognition engine and that language’s data, about 4 to 6 MB in total, are downloaded from this site and saved by your browser. That download only brings files to your device and sends no part of your document. After that, the tool works offline for that language. The Prove it’s private panel under the tool counts what the page sends while you work.
If it doesn’t work
- “No text could be recognized.” Check the language, and that the page is upright and sharp. A page that is mostly a photo or a drawing may have no text to read.
- Lots of wrong words. The language is probably wrong, or the scan is small or blurry. Rescan at about 300 DPI, or take the photo straight on in good light. Turn sideways pages first with Rotate PDF.
- It fails offline. Each language is saved only after its first use. Run one page while online, then it works offline.
- Over 50 pages. Split the PDF and run the parts one at a time.
- A password-protected PDF. Unlock it first with Unlock PDF, using the password you have.
- A signed PDF. Adding text changes the file, so a digital signature in it will no longer be valid. Keep the signed original.
Related tools
- Make PDF searchable: scanned PDFs and photos to a searchable PDF, text or Word.
- Scan to PDF: crop, straighten and clean up photos of paper, with optional OCR.
- Compress PDF: make a scan smaller and keep it searchable.
- PDF to Word explained: what a Word copy can and can’t keep.
- How to redact a PDF: black out text, with an option to keep the rest searchable.