PDF🔒 Runs in your browser

Text Extractor (OCR)

Extract text from scanned PDFs and images using OCR (Optical Character Recognition)

📝

Drop a PDF here or click to browse

.pdf, .png, .jpg, .jpeg, .webp, .tiff, .bmp

Use cases

Digitise a stack of old paper invoices for accounting search and retention.
Make a scanned textbook full-text searchable for studying.
Extract text from a scanned contract to copy-paste a clause into a draft.
Prepare a scanned document for translation by extracting its text first.
Convert receipt scans to text fields before pasting into an expense-tracking spreadsheet.

About Text Extractor (OCR)

PDF OCR turns scanned, image-based PDFs into searchable, copy-pastable text using Tesseract.js, the open-source OCR engine — running entirely in your browser. Drop a scanned document, pick the language (40+ supported), and watch the engine recognise text page by page. The result can be exported as plain text or as a 'searchable PDF' that keeps the original page images but adds an invisible text layer. Use it to digitise old paperwork, make a scanned book full-text searchable, extract content from receipts for an expense tool, or prepare a non-editable PDF for translation.

Frequently Asked Questions

Which languages are supported?

Over 100 languages are supported through Tesseract.js, including English, French, Spanish, German, Chinese, Japanese, Arabic and many more. You can also run multi-language OCR by selecting several languages at once.

Is my PDF uploaded to a server?

No. Text recognition happens entirely in your browser via WebAssembly. Your files never leave your device.

What output formats are available?

The recognized text is available as plain text (.txt). You can copy it or download it once OCR finishes.

Why does OCR sometimes mis-recognise letters?

OCR accuracy depends heavily on the source quality. Crisp 300 DPI scans of printed text produce 95–99 % accuracy. Low-resolution photos, handwritten text, decorative fonts, or pages with stamps/annotations crossing the text drop accuracy significantly. Choosing the correct language matters too — running French text through an English-only model multiplies errors.

Why is OCR slower than expected?

Tesseract.js loads a 4–10 MB language model on first use and runs computer-vision passes per page in the browser's main thread. Expect ~3–8 seconds per page on a modern laptop, longer on phones or for large documents. The progress bar updates live so you can plan accordingly. Subsequent pages are faster because the model stays in memory.

Use cases

Digitise a stack of old paper invoices for accounting search and retention.

Make a scanned textbook full-text searchable for studying.

Extract text from a scanned contract to copy-paste a clause into a draft.

Prepare a scanned document for translation by extracting its text first.

Convert receipt scans to text fields before pasting into an expense-tracking spreadsheet.