Make a Scanned PDF Searchable

PDF (Portable Document Format) preserves exact layout, fonts, and graphics across every device, which makes it the standard for sharing finished documents. File Converts turns your Scanned PDF into Searchable PDF right in your browser — no upload, no sign-up, and no daily limit. Because the conversion runs on your own device, even sensitive files stay private, and it keeps working offline once the engine has downloaded on your first use.

Pick the language before you start

The language menu offers English, Spanish, French, German, Italian, Portuguese and Dutch. Each run reads one language, and choosing the right one makes a real difference: the recognizer uses it to decide what words and letters to expect. For a document that mixes languages, pick the one most of the text is in.

The text layer can only hold the Western European (Latin-1) character set. Accented letters such as é, ñ, ü and ß are fine. Letters outside that set, such as Polish ł or Czech ř, are dropped from the layer, which is why those languages are not offered. A word containing one would be stored misspelled and could not be found.

How to read the confidence score

Every recognized word comes with a confidence score. Words below 40% are left out of the text layer entirely. A smudge read as nonsense is more harmful in a search index than a gap, because it makes the document look searched when it was not. The figure shown after recognition is the average over the words that were kept.

If that average is under 75%, the tool warns you to expect mistakes. Open "Read what was recognized" and skim it. It shows the first few thousand characters, which is enough to tell a clean read from a bad one. Names, figures and reference numbers are worth checking against the page, because those are what people search for.

What changes in the file, and what doesn't

The page images are copied through untouched. Nothing is re-drawn or re-compressed, and the document's properties (title, author, dates) are left as they were. What is added is an invisible word for every recognized word, placed over the ink and stretched to the same width. Selecting a word highlights that word, not a block drifting off to the side.

Some characters are normalised as they go in. Curly quotes become straight, en and em dashes become hyphens, an ellipsis becomes three dots, and fi and fl ligatures are split into separate letters. As a result, a search typed on an ordinary keyboard finds them. Pages are read at 144 pixels per inch, so very small print on a low-resolution scan may not be picked up.

PDFs that already have some text

The tool does not check whether a page already has selectable text. It recognizes every page. Run on a document that was born digital, it adds a second invisible copy of the words over the real ones, so search finds every match twice and copied text comes out doubled. Before you start, try selecting a sentence. If it highlights, the page does not need OCR.

For a mixed file, such as a typed report with a few scanned signature pages, split the scanned pages out with Split PDF, make those searchable here, and merge everything back with Merge PDF.

Large scans, memory and time

Files can be up to 100 MB. If a file is large compared with the memory your browser reports, you get a warning before you start. It may still work, but a phone is where a long scan is most likely to run out of room. Pages are processed one at a time and each one is released before the next, with a percentage on the button and a Cancel button beside it.

The first run downloads the recognition engine and its language data, which takes a moment. After that it is kept on your device and later runs start straight away, even offline. If you plan to work without a connection, run one file in each language you need while you are still online.

What to do when no text is recognized

If the result says no text was recognized, the usual causes are blank pages, pages that are sideways or upside down, or a scan too low in resolution to read. The tool does not straighten or rotate pages for you. Rescanning the document upright, ideally at 300 DPI or more, fixes most of these. Photos of pages taken at an angle or in poor light read badly for the same reason.

If it stops with "Recognition failed", the PDF is usually damaged or password-protected. Remove a password with Unlock PDF, or re-save a damaged file from a PDF viewer, and run it again.

Recognizing text without uploading

Recognition runs in your browser and the scan never leaves your device. That matters because the documents people make searchable are usually archives of signed contracts, medical letters and old statements. The output downloads with -searchable added to the name, and it can go straight on to PDF to Text or Compress PDF without being chosen again.

What this tool does

How to convert Scanned PDF → Searchable PDF

Frequently asked questions

Related conversions