The tools on this page need JavaScript. Your browser has it turned off, so the tools will not respond. Your files go nowhere: they are processed by your browser itself. Without JavaScript, there is nobody to do the work. That is the direct trade-off of what makes this site. The rest of the page reads normally.

Skip to content

Make a scan searchable without sending it anywhere

zero upload

Drop your scan here

a PDF, or photos of pages

Zero size limit · the text is read on your device

The fifty-seven other tools

popularCompress a PDFShrink the file without harming the text. Merge PDFsCombine several documents into one. Split a PDFCut a document into several files. Extract pages from a PDFKeep only the pages you want. Delete pages from a PDFRemove the pages you no longer want. Rearrange a PDFChange the page order with your mouse. Rotate a PDFStraighten pages scanned crooked. Number PDF pagesAdd continuous page numbers. Watermark a PDFLay a word across every page. Create a fillable PDFYour document becomes a form. Sign a PDFYour signature, placed wherever you want. Redact a PDFHidden text is destroyed, not covered. Crop a PDFCut the white margins off a scan. Multiple pages per sheetTwo, four or nine pages on one sheet. Remove PDF metadataAuthor, software and dates removed from the file. Unlock a PDFRemove a password you know. Edit a PDFText, shapes and images on your pages. Add a header and footer to a PDFNumbers, date and Bates in the margins. Compare two PDFsThe differences between two versions. Repair a PDFRecover a damaged PDF. PDF to imagesEach page becomes a JPG or a PNG. Images to PDFYour photos become one document. PDF to textA PDF’s text as .txt, .md or .html. Extract images from a PDFPull out the photos stored in a document. PowerPoint to PDFOne slide per page, text stays text. Word to PDFThe content comes through, and so does the page layout. EPUB to PDFAn e-book becomes a PDF. Create a zipGather your files into one archive. Password-protect a zipAn archive encrypted with AES-256. Excel to CSVOne .csv per sheet, accents preserved. Convert a 3D modelBetween .glb, .obj and .stl, losses stated up front. popularCompress an imageLighten your photos without spoiling them. Convert an imageSwitch to JPG, PNG or WebP. Resize an imageIn pixels or as a percentage. Rotate an imageStraighten or flip a photo. Crop an imageCrop to the frame and format you want. Watermark an imageA text repeated across the whole image. Remove EXIF from a photoGPS and camera removed, pixels untouched. HEIC to JPGThe iPhone photos Windows won’t open. WebP to JPGThe format everything opens. WebP to PNGTo keep transparency. PNG to JPGTo lighten it when size gets in the way. JPG to PNGUseful before editing, not after. JPG to WebP25 to 35% smaller. PNG to WebPMuch lighter, transparency kept. Open a Word fileRead a .docx without installing Word. Open an Excel fileRead a workbook sheet by sheet. Open a PowerPointFlip through a .pptx and its notes. Open an EPUB bookRead an .epub without an e-reader. Unzip a fileOne archive or a hundred, a single zip out. Open a 3D modelView a .glb, .gltf, .obj or .stl. QR code generatorA code computed on your device that never expires. Password generatorA verified draw, an exact entropy. Username generatorUsernames you can pronounce. Voice recorderYour voice, recorded on your device. Record your screenRecord your screen as MP4. Record your cameraYour webcam, without sending anything.

How it works

A scanned document is just a series of images: you can see the text, but you can’t search it, copy it or have it read aloud. The tool draws each page, recognizes the words, then places an invisible text layer on top, word by word, exactly where each word is drawn. The document keeps its look down to the pixel, and becomes searchable.

Recognition is done by Tesseract, the reference open-source engine, installed in your browser for the time of the job. The engine and the chosen language download on first use, 4 to 7 MB depending on the language, then stay in the cache.

What recognition reads well, and less well

Printed text, scanned straight and sharp, reads almost without mistakes. Fine reading, at 300 dots per inch, helps with small print and footnotes. A tilted, blurry or very high-contrast scan gives errors, as in any OCR software.

Handwriting isn’t recognized. Tables are read line by line: their text becomes searchable, but the column structure isn’t rebuilt.

Why tick the right language

The engine relies on a model specific to each language, which knows its accented letters and common words. A French text read with the English model loses its accents. For a document that mixes two languages, tick both: reading is a little slower.

Frequently asked questions

Is my file sent anywhere? No. You can check for yourself in thirty seconds: here is how.

Does the document look different? No. The pages keep their original image, with no recompression. The added text layer is invisible: it is there to search, select and copy.

What about pages that already have text? They’re left as they are, so that every word isn’t found twice in a search. A checkbox lets you read them anyway, for example if their text is unreadable.

How long does it take? A few seconds per page on a recent computer, a little more on a phone. Fine reading takes about twice as long as fast reading.

What about password-protected PDFs? A PDF that opens without a password but blocks printing or copying is accepted as it is. If it takes a password to open, go through unlock a PDF first, if you know it.