Scanned PDF or photo to text (OCR)
Extract printed English or Hindi text from a scan, photo or PDF. Review and correct the result, then save text or Word, or download a searchable PDF. Handwriting, tables and complex layouts may not be recognised reliably.
One PDF (up to 10 pages) or JPG, PNG, WebP · 20 MB maximum · image limit 16 million pixels
First use loads the OCR engine and language model from this website (several MB). Large pages may be slow on phones. Your file contents stay on your device.
OCR can misread numbers and words. Check against the original. Exports contain extracted text; original document layout is not preserved.
Searchable PDF: keeps each scanned page as an image and adds invisible recognised text, so you can search, select and copy it. It is a standard PDF, not PDF/A. It is a rasterized copy, so interactive form fields and digital signatures from the original are not kept. Your edits in the box above apply to TXT, Word and copy only; the searchable PDF uses the original recognised text.
How to get text from a scan
- Add your scan. Choose one PDF of up to 10 pages, or a JPG, PNG or WebP photo of a printed page.
- Pick the language and recognise. Select English, Hindi or English + Hindi to match the printed text, then select Recognise text. Each page is processed in turn, and you can cancel at any time.
- Check, then save. Correct any misread words in the text box and save TXT or Word, or download a searchable PDF of the original pages.
Worked example: the two-page sample
Select Try scanned PDF. PixnPage creates a two-page scanned PDF on your device, with the lines “PIXNPAGE SCANNED SAMPLE”, “Invoice number: 2026-104” and “The library opens at 9 am on Saturday.” Leave the language on English and select Recognise text.
The text box shows each page under a “--- Page 1 ---” or “--- Page 2 ---” marker, and the status line reports the average recognition confidence. Select Download searchable PDF, open it in your PDF viewer and search for 2026-104: it is found on both pages, while each page still looks like the original scan.
Formats, limits and privacy
- Input: one PDF (up to 10 pages) or a JPG, PNG or WebP image.
- Limits: 20 MB per file; images up to 16 million pixels.
- Languages: printed English, Hindi (Devanagari) or both together.
- Output: plain text (.txt), a text-only Word file (.docx), or a searchable PDF. The searchable PDF is a standard PDF, not PDF/A.
- Privacy: recognition runs in your browser. The OCR engine and language data load from this website on first use (several MB); your file is not uploaded.
Frequently asked questions
Can it read handwriting, tables or multi-column pages?
Not reliably. The OCR is designed for clear printed text. Handwriting is usually misread, and tables, columns and complex layouts may come out in the wrong order. TXT and Word exports contain the recognised text only, without the original layout.
If I correct the text, does the searchable PDF change too?
No. Your edits apply to Copy, TXT and Word only. The searchable PDF keeps each page image with the invisible text layer exactly as it was first recognised. To improve that layer, try a sharper scan and run OCR again.
Is the searchable PDF the same as my original PDF?
It looks the same but is a new, rasterized copy: every page is stored as an image with hidden text added. Interactive form fields, links and digital signatures from the original are not kept, and it is not a PDF/A archive file. Keep your original.
Why is the recognised text wrong or empty?
Common causes are a blurred, tilted or low-contrast scan, or the wrong language setting (for example, Hindi text with English selected). Try a straighter, sharper scan, choose the matching language, and always check numbers and names against the original.
My PDF already has selectable text. Do I need OCR?
Usually not. If you can already select words in the PDF, PDF to text / Word extracts that text directly, without recognition errors.
Related tools
Using Scan to Text (OCR)
Choose a scan, select a language, recognise text, correct errors and save the result.
Example: Extract text from a photographed receipt
- Use a straight, sharp photo with good lighting. Choose the correct recognition language.
- Run recognition and compare the extracted text against the receipt, especially totals, dates and decimal points.
- Correct any recognition errors before saving text or a searchable PDF. OCR is not guaranteed accurate and the output is not PDF/A certification.
What to check before sharing
OCR can misread words. Review names and numbers. Searchable PDF is not certified PDF/A.