OCR: Turn Scanned PDFs and Images Into Text
Get editable text out of scanned documents, photos of pages and image-only PDFs. PDFSpark runs optical character recognition in your browser, page by page, in 12 languages. Copy the text, download it as a .txt file, or save a searchable PDF. Your scans are never uploaded to an OCR server.
Your file is processed by your own browser and never uploaded. Most online tools send your document to their servers and delete it later — this one never receives it at all.
Why use OCR PDF
- Works on scanned PDFs and JPG, PNG or WebP images
- 12 languages, including English, Spanish, French, German, Chinese, Japanese, Arabic, Hindi and Russian
- Searchable PDF output with an invisible text layer
- Copy to clipboard or download as .txt
- Page-by-page progress, with a Cancel button
- OCR runs locally with Tesseract; nothing is uploaded
Popular uses
- Copy text from a scanned contract or letter into an email or document
- Pull figures from a photographed receipt or invoice
- Recover the text of an old printed report that only exists as a scan
OCR PDF: frequently asked questions
What is OCR?
OCR, or optical character recognition, is technology that reads the letters in an image and turns them into real text you can copy, search and edit. Scanned PDFs are just pictures of pages, so without OCR you can't select their words. PDFSpark uses the open-source Tesseract engine running in your browser.
How do I extract text from a scanned PDF?
Open the scan in PDFSpark's OCR tool, pick the language, and click Recognise text. Each page is rendered and recognised in your browser, and the combined text appears with buttons to copy it or download it. For PDFs that already contain selectable text, PDF to Text is faster.
How accurate is OCR?
Clean, high-contrast scans of printed text usually come out very well. Accuracy drops with blurry photos, skewed or low-resolution scans, unusual fonts, handwriting, and complex layouts such as tables or multiple columns. Scanning at 300 DPI, keeping pages straight and choosing the correct language all help. Always proofread important numbers.
Can PDFSpark make a searchable PDF?
Yes. After recognition you can download a searchable PDF: each page keeps its scanned look, with an invisible text layer so you can search, select and copy words. For PDF input the pages are stored as 150 DPI images.
Which languages are supported?
English, Spanish, French, German, Italian, Portuguese, Dutch, Simplified Chinese, Japanese, Arabic, Hindi and Russian. Choose the language that matches your document before you start, because the recognizer uses language-specific data to read accents and characters correctly. That data is downloaded to your browser the first time you use it; your document is not sent anywhere.
Is it safe to OCR confidential documents?
With PDFSpark, yes. The OCR engine runs inside your browser, so scanned pages are recognized on your own device instead of on a remote server. The only things downloaded are the recognition engine and language data; your document and its extracted text never leave your computer or phone.