YOUR DOCUMENTS, ON YOUR DEVICE
Private OCR for Scanned PDFs and Photos
Recognize printed text in scans and photos on your device. Choose the language, review the recognized text and download your result.
- Add a scan or PDF
- Recognize and review text
- Download your result
Your document stays on this device. No account or file upload needed. How privacy works · All PDF tools
Text edits update the TXT download only. The searchable PDF keeps the original OCR text layer.
How to recognize text
- Choose your scanned PDF or document photo.
- Select the document’s language and, for a PDF, the pages to recognize.
- Choose whether to create a searchable PDF, then select Recognize text.
- Compare the recognized text with your original and correct any errors.
- Copy the text or download your TXT or searchable PDF.
Make paper text useful again
Copy a passage from a scanned handout into your notes, find a word in a receipt scan, or make an image-based document searchable. Start with a sharp, upright source: recognition cannot restore letters hidden by blur or glare.
For a document that already has selectable text, PDF to Text is quicker and avoids recognition errors. Use Document Scanner first if a paper photo needs straightening.
Frequently asked questions
Can this read a scanned PDF or a photo?
Yes. OCR looks at page images and attempts to recognize printed text. Choose one PDF, JPG, PNG or WebP image. For PDFs, you can select up to 10 pages per run.
Are my documents sent to an OCR server?
No. Recognition runs in a WebAssembly worker on your device. Kalika serves the engine and selected language files, but the tool does not upload your document or recognized text. It does not save documents or results in browser storage. Ordinary browser caching may retain public engine and language assets.
Which recognition languages are available?
Choose English, French, Spanish, Hindi, Nepali with English, or Simplified Chinese. Select the language used in the document; the website interface language does not choose it for you.
How accurate is the recognized text?
Check every result, especially names, numbers and accents. Clear, upright printed text works better than handwriting, low-resolution photos, shadows, curved pages or complex tables. Reading order and formatting may be wrong. This is not a handwriting transcription service.
What changes in the searchable PDF?
The output contains rendered page images with an OCR text layer. It is a new copy of the selected pages, not a preservation of the original PDF structure. Links, bookmarks, forms, accessibility tags and digital signature validity are not retained. Edits in the text box affect TXT downloads only, not the searchable PDF.
Why does OCR take time on my phone?
The engine and selected language files load when you start recognition. Pages are processed sequentially and working images are capped at 2,200 pixels on the longest side. Try one page first on a phone; use Cancel if it takes too long. Large or faint documents may need a clearer source image.