Online OCR
Online OCR turns the words in a picture or a scanned page into text you can copy, search and edit. Upload an image or PDF, choose the language, and get plain text, a Word file, a searchable PDF or Markdown back.
Image to Text
or drop PDF, JPG, JPEG … files here ·
Max 95 MB per file· Up to 20 files· Deleted after 30 minutes
Files are processed on our servers and deleted after 30 minutes.
For pages that mix two languages, e.g. Hindi with English words. Leave on none otherwise; an extra language slightly lowers accuracy.
Preserve keeps columns, paragraphs and tables where they were. Plain outputs one continuous block of text, easier to paste into forms.
PDF and multi-page TIFF input: read every page, only the first, or a range.
Pages to read, e.g. 1-3, 5, 8-10.
How to use this tool
- 1
Upload images or scans
Add photos, screenshots, scanned PDFs or multi-page TIFFs, up to 20 files and 95 MB each.
- 2
Set language and output
Pick the Document language and an Output: Plain text (.txt) to copy, Word (.docx) to edit, Searchable PDF to keep the scan, or Markdown.
- 3
Recognize and copy
Press Recognize text, open the result and copy or save what you need.
OCR reads printed text very well and everything else less well: how the page was captured decides accuracy more than the software does.
What online OCR does with your file
Optical character recognition is older than the personal computer: the Kurzweil Reading Machine read printed books aloud to blind users in 1976. The open-source Tesseract engine began at Hewlett-Packard in the 1980s, was released as open source in 2005 and gained a neural-network (LSTM) recognizer in version 4.0 in 2018. Modern engines find text regions, then lines, and recognize each line as a sequence rather than letter by letter, which is why they cope with tight spacing and ligatures better than older software.
What comes out is text plus positions. Choose Plain text (.txt) when you want to paste the words somewhere, Word (.docx) to edit with the layout kept, Searchable PDF when the scan must stay as it is, and Markdown for notes and documentation. Students run it on past papers and printed handouts, offices on invoices and letters, developers on screenshots of error messages.
- files per batch
- 20
- per file
- 95 MB
- then deleted
- 30 min
- outputs: TXT, DOCX, PDF, MD
- 4
What decides OCR accuracy
| Do | Don't |
|---|---|
| Scan or photograph at 300 DPI; text smaller than 10 pt needs it most | Upscale a small, blurry image; enlarging adds no detail |
| Shoot square-on in even light, so the page is not a trapezoid with a shadow | Trust numbers without checking; 5 and S or 0 and O are easily swapped |
| Keep dark text on a light, plain background | Photograph glossy paper under a lamp or flash; glare erases letters |
| Choose the right Document language, plus Also recognise for mixed pages | Expect cursive handwriting or decorative fonts to come out clean |
| Crop away neighboring pages, fingers and the desk | Turn on Enhance contrast for clean scans; it is meant for dim photos |
What OCR cannot do
OCR recognizes characters, not meaning. It cannot be sure whether a line is a heading or a caption, it cannot read text hidden under a stamp, and it has no idea whether a misread number makes sense. Charts, diagrams and photos are not described; only words printed inside them are read. Mathematical notation comes out as plain characters, not as equations.
It is not a translator either: the result is in the language of the original. For a document that already contains real text, OCR is the wrong tool, since PDF to text reads the existing text faster and without errors. For scans that should stay PDFs, use OCR PDF, and for Word output from pictures, image to Word.
Online OCR questions
Is this online OCR free to use?
Yes. There is no account and no payment step, and each batch takes up to 20 files of up to 95 MB. Results are ready to download as soon as recognition finishes.
Which file types can I upload for OCR?
JPG, JPEG and JFIF, PNG, WebP, HEIC and HEIF photos from iPhones, TIFF and multi-page TIFF scans, BMP, GIF and PDF. Each file is recognized and saved in the output you choose.
How does this compare with Google Docs or Apple Live Text?
Google Drive can open an image or PDF as a Google Doc with OCR, though Google's help asks for files of 2 MB or less, and Apple's Live Text, added in iOS 15, copies text from photos on the device. Both are fine for a quick copy. A dedicated OCR tool adds language choice, page ranges, batches and output as DOCX, searchable PDF or Markdown.
Can OCR read handwriting?
Set Text type to Handwriting for notes and forms filled in by hand. Neat print-style writing reads reasonably well; connected cursive, crowded margins and faint pencil are hard for any engine, so plan to correct the result.
Does OCR work on a screenshot of a phone screen?
Very well, because screen text is sharp and level. Crop out the status bar and keyboard first so they do not end up in the text.
Are my documents stored?
Uploads and results are deleted automatically 30 minutes after processing, and no account is involved. The privacy policy has the details.
- OCR PDFAdd a searchable, selectable text layer to scanned PDFs, or pull the text out to Word or TXT. 34 languages, page ranges, 20 files at once, free.
- PNG to Text ConverterConvert PNG to text: read the words in a PNG image, UI mockup or infographic and save them as a .txt file to paste anywhere. 34 languages, 20 images at once.
- Screenshot to TextTurn screenshots into copyable text: error messages to search, dashboard tables to paste, chat logs to quote. Up to 20 screenshots at once, 34 languages.
- Handwriting to Text ConverterConvert photos and scans of handwritten notes, forms and letters to text you can copy or edit. What print, block capitals and cursive each give you.
- JPG to Word ConverterTurn a JPG photo of a printed page into an editable Word document. OCR in 34 languages keeps paragraphs and tables; up to 20 images at once, free.
- Hindi OCR: Image to Hindi TextHindi OCR for photos, scans and PDFs: recognize Devanagari text, matras and conjuncts included, as Unicode you can copy, edit and search. Hindi + English pages.