How to OCR a PDF

Reading a scanned PDF as text

A scanned PDF is a picture of a document, not a document. You can read it on screen, but you cannot search it, copy a paragraph from it, or paste it into another file. OCR — optical character recognition — is the step that turns the picture back into text: it looks at every page, recognizes the letters, and writes them out as text you can use. This tool runs OCR in your browser, with the Tesseract engine.

Each page is read in order, and the result is a .txt file with the text of every page, with page markers between the pages so you always know where one page ends and the next begins. The result opens in any text editor, and from there you can search it, copy from it, or feed it into another tool — a summary, a translation, or a search. One file per run, up to 100 MB and up to 20 pages. The first time you use the tool, the OCR engine (about 7 MB) and the language file for your document (about 1–4 MB) download into your browser; after that they are cached and ready.

The honest limit is that OCR guesses. It looks at pixels and decides which letter each shape is, and on a blurry, slanted, or faded page it will guess wrong. A clean scan at a good resolution reads well; a photo of a page taken at an angle, in bad light, reads badly. Always check the result — especially names and numbers, where a single wrong letter changes the meaning. Only printed text is supported; handwriting is not, and no setting will make it work. And choose the language of the document before you start: OCR with the wrong language file will misread letters that look similar across languages.

A few practical tips. If you can, scan instead of photographing — a flat scan is the best input OCR can get. If the scan is skewed, straighten it first; a tilted page is the most common reason for bad results. Run a short test on one page before committing a whole document, so you know what quality to expect. And when the text matters — a contract, a medical record, a set of figures — read the .txt file against the scan, line by line, before you rely on it. OCR saves hours of typing, but it does not remove the need to check.

It also helps to know what kind of document OCR handles well. A clean printed page — a book, a form, a typed letter — reads well, because the letters are regular and the spacing is even. A page with two columns, a table, or a mix of text and pictures reads less well: the text comes out, but the order of the columns can get mixed, and a table can come out as one long line of words. A page with a stamp, a signature, or handwriting in the margin will read the printed text and skip the rest, because handwriting is not supported. If the layout matters — a form where each field must stay in its place — keep the scan alongside the text, so you can check where each line came from.

The language choice deserves a second look, too. OCR reads best when the language file matches the language of the page, because letters that look alike across languages — a German umlaut, a French accent, an Indonesian word with no English twin — are guessed from the language you chose. A document that mixes two languages will read one of them well and the other badly, and there is no setting that reads both at once. If your document is mostly one language with a few words in another, pick the main language and check the foreign words by hand.

Everything runs in your browser. Your scan is never uploaded to a server, so a personal document stays on your device. There is no account, no daily limit, and no watermark. Open the tool, drop your scan, pick the language, and get the text back.

How to use this tool

1. 1Drop one scanned PDF onto the OCR PDF page — up to 100 MB, up to 20 pages per run.
2. 2Choose the language of the document.
3. 3Press Read text with OCR and wait while each page is read in order.
4. 4Download the .txt file and open it in any text editor.
5. 5Check the result carefully — especially names and numbers — before you rely on it.