Searchable PDF (OCR)
A scan or a photo of a document is only a picture: you cannot search for a word in it. Text recognition (OCR) reads your pages and slips the text under the image. The PDF keeps its look and becomes searchable, copyable and indexable. Everything happens on your device.
Première fois ? Le déroulé, pas à pas
- Appuyez sur « Choisir des PDF ou des photos », ou glissez vos fichiers sur la page.
- Cochez la langue du document. Avec plusieurs fichiers, choisissez entre « Un PDF par fichier » et « Un seul PDF réunissant tout ».
- Appuyez sur « Reconnaître le texte » et laissez la lecture se faire.
- Appuyez sur « Télécharger le PDF » pour chaque document, ou sur « Tout télécharger (.zip) » s'il y en a plusieurs.
Conseils
- Pour un bon résultat, scannez à 300 points par pouce, en noir sur blanc, bien à plat.
- Ne cochez que les langues présentes dans le document : chaque langue en plus ralentit la lecture.
- Regardez la confiance affichée pour chaque document : si le scan est signalé peu lisible, relisez le texte avec « Voir le texte reconnu ».
- Sur téléphone, la lecture est plus lente que sur un ordinateur récent : traitez peu de pages à la fois.
1Choose your scans
or drop them here: scanned PDFs, photos of documents (JPG, PNG, WebP).
Your documents are read on your device: they are neither sent nor stored. The recognition engine is downloaded from this site on first use (about 5 MB), then kept by your browser.
2Get your searchable PDFs
Good to know
What is a searchable PDF for?
A PDF that comes out of a scanner is a series of photos: Ctrl+F finds nothing in it, and the computer does not know what it contains. Once the text is recognised, you can search for a word in the document, copy a passage, and the search of Windows, macOS, your mailbox or your document management tool finds the file by its content.
Are my documents sent anywhere?
No. The reading is done by your browser, on your device, with the open-source Tesseract engine. The engine and the languages come from our site, not from a third-party service. Once loaded, you can even go offline: nothing leaves your device.
Does my PDF look different afterwards?
No. Each page remains the original image; the recognised text is written over it in invisible ink, word by word, exactly where it sits. When you select a line, it is this hidden text that you copy. Pages that already contained text are left as they are.
Is the recognised text reliable?
On a sharp, straight scan, about 95 words out of 100 are right, often more. The confidence shown for each document gives you an idea. Quality drops with blurred or skewed photos, small print, stamps and coloured backgrounds. Handwriting is not recognised. For a good result: scan at 300 dots per inch, black on white, nice and flat.
How long does it take?
A few seconds per page on a recent computer, longer on a phone. Several pages are read at the same time. The first time, the engine is downloaded (about 5 MB with one language); it is then kept by your browser. You can keep working in another tab while it reads.
What about technical drawings?
Title blocks and notes read well. Tags placed in the middle of lines, rotated or very small text are recognised less well: count on a search aid, not on an exact survey. To annotate the drawing afterwards, open the resulting PDF in the PDF editor.
Which characters are recognised?
French, German and English, with their accents and special characters (é, ç, œ, ä, ö, ü, ß, €). Other alphabets (Greek, Cyrillic, Arabic…) are not supported.