Saved

Press the star next to a tool's name to save it.

All my tools

OCR PDF: Make a Scanned PDF Searchable

Use this OCR PDF tool to make a scanned PDF searchable: it recognises the words on each page and adds them as an invisible text layer, so you can search and select the text. It runs in your browser with no upload, in English.

This reads English text. The first run downloads the recognition engine and language data (about 8 MB) from this site, which can take a moment. Each page is recognised at the resolution you choose, and up to 30 pages are read per run.

Your PDF is read on this device and is never uploaded. The recognition engine loads from this site.

Your searchable PDF

Choose a scanned PDF and press Make searchable.

What this OCR PDF tool does

A scanned PDF is a set of pictures of pages. You can see the words, but the file holds no text, so you cannot search, select or copy it. Optical character recognition, or OCR, converts pictures of printed text into text a computer can use. [1] This OCR PDF tool recognises the words on each page and adds them to the PDF as an invisible layer on top of the page picture.

The result looks exactly like your scan. The difference is that search and copy work, and screen readers can read it. Many people call this a searchable PDF. [1]

How the recognition works

The tool draws each page with PDF.js, [4] then reads it with Tesseract, an open-source recognition engine released under the Apache License. [2] Version 4 and later use an LSTM neural network. [2] The language data comes from the “fast” English models. [3] Everything runs in your browser, including the engine, which is served from this site.

How to make a scanned PDF searchable

  1. Choose your scanned PDF.
  2. Pick a resolution. 200 dpi suits clean scans, and 300 dpi helps small print.
  3. Press Make searchable. The first time, the engine takes a moment to load.
  4. Wait while each page is read. A progress line shows where it is.
  5. Download the new PDF and try searching it.

What you will get

  • A PDF that looks the same as your original, with a hidden text layer behind each page.
  • The recognised words placed to match their position on the page, so selecting and highlighting land in the right place.
  • The word count found, in the result line.

The words are written with a basic Western font, so only Western letters, numbers and common punctuation are stored. Other characters are dropped.

Getting better recognition

  • Use clear, upright scans. Skewed or rotated pages recognise poorly.
  • Raise the resolution to 300 dpi for small print.
  • Dark, noisy or low-contrast scans lose words. Rescan with more contrast if you can.
  • Handwriting and decorative fonts are not recognised well.
  • Only English is supported. Other languages come out as nonsense.

Limits

  • Up to 30 pages are read per run. Split a longer PDF first.
  • Pages that carry a rotation setting are skipped, and a note tells you how many.
  • Recognition is not perfect, so check important words, numbers and names.
  • The file may grow a little because of the added text.

A worked example

You scanned a 6-page letter on a copier and received “scan.pdf”. Searching for a name in a viewer finds nothing, because the file holds only pictures. You choose the file, pick 200 dpi and press Make searchable. The tool draws each page, reads the words and adds them to the PDF. The result line might say “6 pages · 1,240 words found”. In the new PDF, searching for the name highlights it on the page, and you can select a paragraph and copy it into an email. [1]

The page pictures themselves are not changed, so the document looks exactly as scanned. If a word was misread, for instance “rn” read as “m”, the picture still shows the right word, but searching for it may fail. Check key names and numbers.

Common mistakes

  • Using a PDF that already has text. It will add a second layer. Test by trying to select text in your viewer first.
  • Scanning crooked pages. Straighten them first, because slanted lines recognise poorly.
  • Choosing a low resolution for small print. Use 300 dpi if words are missed.
  • Trusting the result without checking. Recognition is good on clean print and weaker on noise, stamps and handwriting.
  • Running a non-English document. This tool reads English only.

Troubleshooting

  • If it says no text could be recognised, try a higher resolution or a cleaner scan.
  • If the first run is slow, wait. The engine downloads once and is cached by your browser.
  • If the tab slows down, split a long PDF into smaller files first. Only 30 pages are read per run.
  • If some pages were skipped as rotated, rotate them in the Rotate PDF tool and try again.

A searchable PDF is also easier for assistive software to read, which is one reason offices ask for them. [1]

What happens to my document

Your document is not edited visually. The tool keeps the original PDF and draws the recognised words on top of each page, using a font that is invisible to the eye. A viewer that supports text search will find those words and highlight them where they appear. Printing the PDF prints the pictures, not the hidden text, so your printed copy looks just as it did before.

Hidden text is stored as ordinary PDF text commands. [4] Because we fit each word to the width recognised on the page, the highlighted area lines up with the word. Mistakes in recognition show up as missing or wrong search hits, so it is worth searching for a few words you know are on the page.

Privacy

The engine, the language data and your PDF all stay in your browser. Nothing you choose is sent to a server.

What next

Use this often? Add OCR PDF: Mak to your Home screen.

Frequently asked questions

How do I make a scanned PDF searchable?

Choose the scanned PDF, pick a resolution and press Make searchable. The tool recognises the words and adds an invisible text layer, then you download the new PDF.

What does OCR PDF do?

OCR turns pictures of printed text into real text. The tool adds that text invisibly on top of each page, so you can search, select and copy it.

Which languages are supported?

English only. Other languages will not be recognised correctly.

Why is the first run slow?

The recognition engine and the English language data, about 8 MB, load once from this site. Later runs in the same session are faster.

Is my PDF uploaded?

No. The engine runs in your browser and your PDF never leaves your device.

What went wrong?

We only receive the reason you pick, never what you typed into the tool.

Thanks for your feedback.

More PDF Tools