PDF Tools

OCR PDF

Reads text out of scanned PDFs and images using Tesseract.js OCR running in your browser. You get the recognised text as a TXT download and, for English, a searchable PDF built from the page images with an invisible text layer.

Try in Workspace
Runs in your browser with Tesseract.js — the file is not uploaded. Text recognition is only as good as the scan: clear, upright, 200–300 DPI pages work best. Language data (a few MB) is downloaded from a public CDN the first time you use a language.

How to Use

  1. 1Click the upload area and choose a scanned PDF or a PNG, JPG or WebP image (up to 100 MB).
  2. 2Pick the language of the text in the Language of the text list: English, Urdu, Arabic or English + Urdu.
  3. 3Click Run OCR. The progress line shows which page is being read; the first run also downloads language data.
  4. 4Read or copy the result from the text box, then click Download TXT.
  5. 5For English results, click Download searchable PDF to get a copy of the pages with selectable text.

About this tool

Updated

How the text is recognised in your browser

OCR turns a picture of text into real characters. Here each PDF page is drawn to a canvas at 2x scale and read by Tesseract.js on your own device, using the language you select.

You choose one input file (PDF, PNG, JPG or WebP, up to 100 MB) and one language: English, Urdu, Arabic or English plus Urdu. The result appears in a text box, with a page marker line between pages of a multi-page PDF, and can be downloaded as TXT. For English you can also download a searchable PDF. It contains the page images with invisible text placed over each recognised line, so you can search and select text in a PDF viewer.

When this fits, and when it does not

Use it for scans and photos of documents that have no selectable text. If your PDF already has real text, OCR is unnecessary.

• A scanned contract or letter that you need to search or quote from. • A photo of a printed page in English or Urdu that you want as editable text. If the PDF already lets you select text, skip OCR and convert it with PDF to Word, which reads existing text but does not do OCR itself. To OCR only a few pages of a long scan, cut them out first with Extract PDF Pages to save time.

Limits and honest caveats

Accuracy depends on scan quality, only one file is handled at a time, and the searchable PDF is English only.

Tesseract is not perfect: handwriting, tilted pages, low resolution and complex tables reduce accuracy, and you will need to proofread. Layout, columns and tables are flattened into lines of plain text. The searchable PDF is rebuilt from JPEG images of the pages, so the original file's vector text and fonts are not retained. Urdu and Arabic are exported as TXT only. The file stays on your device, but language data is downloaded from a public CDN on first use, so an internet connection is needed the first time you use a language. Large multi-page scans are slow because pages are processed sequentially.

Features

  • Accepts one PDF, PNG, JPG or WebP file at a time, up to 100 MB
  • Languages: English, Urdu, Arabic, and English + Urdu
  • PDF pages are rendered at 2x scale before recognition
  • TXT output with a page separator for multi-page files
  • Searchable PDF output for English only: JPEG page images plus an invisible text layer
  • Recognition runs in your browser with Tesseract.js; the file is not uploaded
  • Language data (a few MB) is fetched from a public CDN the first time a language is used

FAQ

Is my file uploaded for OCR?

No. The PDF or image is read and recognised in your browser. The only network request is the one-time download of Tesseract language data from a public CDN, which contains no part of your file.

Why is there no searchable PDF for Urdu or Arabic?

The invisible text layer is drawn with a Latin-only font, so it cannot represent Urdu or Arabic characters. For those languages you get the TXT result only.

How accurate is the result?

It depends on the scan. Clear, upright pages at roughly 200-300 DPI work best. Skewed, blurry, handwritten or very small text will produce mistakes, and nothing is corrected for you, so proofread anything important.

Does the searchable PDF keep the original layout?

It rebuilds each page from a JPEG image of the scan, so the visual look matches, but the original PDF's own text, fonts and vector content are replaced by the image. File size can grow compared with the source.

Is a large scanned document slow?

Pages are recognised one after another on your device, so time grows with page count and computer speed. Files over 100 MB are rejected; split them first.

Related Tools

Extract PDF Pages

Type page numbers or ranges and get a new PDF with only those pages.

PDF to Word

Pull the text out of a PDF into an editable .docx. No OCR, layout or images.

Compress PDF

Re-saves one PDF with metadata fields cleared. Does not recompress images, so savings are small.