or drop a file anywhere here
Files other than PDF are converted first. Signing in raises the size limit.
OCR, optical character recognition, reads the picture of each scanned page and adds its words as hidden text behind the picture, so you can select, search and copy them. This page is for people who hold a scanned PDF, such as a letter, an invoice or a book chapter, and need to search it or copy a passage. Add your PDF, open Make searchable and tick the languages that appear on the pages. The list offers five: Chinese (Simplified), German, English, French and Portuguese. English is ticked to start, and the button stays off until at least one language is ticked. Persian and Arabic are not in the list, so a page in either cannot be read here. The panel says that pages that already have text are left as they are. The result is a PDF, and a signed-out visitor gets it without signing in or signing up, within the free daily limit and the free file size limit. The measured run above shows one real result on a scan we made ourselves.
Our own 2-page A4 scan, image only, English, German and French text, run on indox.org with those three languages ticked
After OCR the words on a scanned page are stored as text under the picture. In our test, searches for harbour, Fische, bateau and 12.5 found them in the output and found nothing in the scan. Use the search box of your PDF reader, or select a line and copy it into another document.
Tick every language that appears on the pages. The panel offers Chinese (Simplified), German, English, French and Portuguese, and shows how many you picked, for example 2 selected. English is on by default. In our test we ticked English, German and French for pages that held all three, and the German and French sentences came out right apart from one accent.
The panel says that pages that already have text are left as they are, so a file that mixes typed pages with scanned ones keeps its typed pages. We also ran a PDF that was already searchable: its text was identical afterwards. The file itself is written again, so its bytes and its creator field change.
A signed-out visitor can run OCR and get the PDF without signing in or signing up. The free daily limit and file size limit apply.
| You want to | Use | What you get |
|---|---|---|
| Search or copy text from a scanned PDF | OCR PDF, this page | The same PDF with a hidden text layer behind each page picture |
| Edit the text in Word | OCR PDF first, then PDF to DOCX | A Word file. On its own, the conversion runs no text recognition on scanned pages |
| Read pages that are sideways | Rotate PDF, then OCR PDF | Upright pages, so the text can be recognized |
Drop one PDF on the upload area at the top of the page, or click to choose it. Try selecting a sentence in your PDF reader first: if nothing highlights, the pages are pictures and OCR is what you need.
Open Make searchable and tick every language that appears on the pages. English is ticked to start. The button stays off if none is ticked.
The button fills with the job's progress and the steps show under it. When it says Done, press Download PDF, or Undo to go back. Search for a word from a scanned page to check.
You get the same PDF with a hidden text layer added to the pages that have none. You choose the languages, English is on to start, and the panel will not run with none picked. We ran the tool ourselves, signed out, on a scan with known text, and the figures in the measured run come from that file. The language list is short, and if yours is not among the five, this tool cannot read it. You do not have to sign in or sign up to get the result, within the free daily limit and the free file size limit.
| What you see | Why | What to do |
|---|---|---|
| Nothing is searchable after the run | The pages already had text and were left as they were, or the text on them is not in a language you ticked | Search for a word from a scanned page. Run OCR again with every language on the pages ticked. |
| Words come out garbled | The scan is low resolution, tilted or faint, or the page is in a language you did not tick | Rescan at a higher resolution, straighten the page and tick every language in the document. |
| An accented letter comes out plain | In our test dégagé was read as degage, so recognition is not exact on small marks | Proofread accented words and fix them in your editor. |
| Persian or Arabic words are not found | Neither language is in the list of five, so those pages were not read | Use this tool only for Chinese (Simplified), German, English, French and Portuguese pages. |
| Sideways pages read as nonsense | The panel has no rotate setting | Turn the pages upright with Rotate PDF, then run OCR again. |
| Make searchable is off | No language is ticked | Tick at least one language. |
| The file changed although it already had text | OCR writes the file again. In our test the text was identical but the file grew by 345 bytes | Keep your original if you need its exact bytes. |
| The signature shows as invalid | OCR writes a new file | Keep the signed original as the record and use the OCR copy only for searching. |
| Mode | Pages that already have text | Pages without text | On this page |
|---|---|---|---|
| skip-text | Copied to the output unchanged | Recognized and given a hidden text layer | Used. The panel shows no mode choice, and in our test a text-only file kept identical text |
| force-ocr | Turned into pictures and recognized again | Recognized and given a hidden text layer | Not offered in the panel |
| redo-ocr | Hidden OCR text from an earlier run is removed and read again | Recognized and given a hidden text layer | Not offered in the panel |
You have scanned articles and book chapters. OCR lets you search them and copy quotes.
Your archive is full of scanned letters and invoices. A text layer makes them findable.
You want to convert a scan to Word. Run OCR first, then convert the searchable PDF.
Facts checked on November 5, 2026
No tool matches that search.
It reads the letters on each scanned page and adds them as a hidden text layer behind the page picture. In our test the picture stayed at its original size, and the words became searchable, selectable and copyable.
The five the panel lists: Chinese (Simplified), German, English, French and Portuguese. Tick each one that appears in your document. With none ticked, the panel asks you to pick at least one and Make searchable stays off.
No. Neither language is in the list, so a page in either cannot be recognized here.
Try to select a sentence in your PDF reader, or search for a word you can see on the page. If nothing highlights or the search finds nothing, the pages are pictures and OCR will give them text.
The panel says that pages that already have text are left as they are. In our test with a PDF that already had text, the extracted text was identical afterwards and no hidden layer was added. The file was still written again, so it grew by 345 bytes and its creator field now names OCRmyPDF.
We measured one run: a two-page scan we made at 200 dpi from known English, German and French text, with a slight tilt and added noise. 0.34% of the characters differed from the known text, none on page 1 and 0.74% on page 2, and the only error was the accent lost in dégagé. That is one clean sample, so treat it as a data point and proofread your own result.
In our test the scan went from 295,431 bytes to 300,465 bytes, 1.7% larger, because a text layer is added and the page picture is kept. That is one file, and we do not promise a size for yours.
No. A signed-out visitor gets the OCR result without signing in or signing up. The free daily limit and file size limit apply.
This tool returns a PDF. Run OCR first, then convert the searchable PDF to Word.
It may. OCR writes a new file, so a signature on the original can stop being valid. Keep the signed original and use the OCR copy for searching.