যেভাবে কাজ করে
How BanglaOCR works
Three steps for you, a Bangla-first AI pipeline underneath — designed around the documents Bangladesh actually produces.
Three steps, under a minute
1
Upload
Drag & drop any scan, photo or PDF — Bangla, English or mixed. Your first 2 pages each day need no sign-up. Pick a page, rotate or crop if needed.
2
AI reads it
The page is read in multiple passes at multiple zoom levels. Printed text and handwriting — Bangla, English or mixed — become editable text.
3
Review & export
Fix anything in the result box, then copy it or download TXT, Word, JSON, HTML or PDF.
Recognition engines
🖨️
Printed Bangla OCR
~99% accuracy on books, certificates, notices and printed forms — GPU-accelerated, seconds per page.
🅰️
English & mixed documents
English OCR at the same grade, and mixed Bangla-English pages — the norm in official paperwork — in a single pass.
✍️
Handwriting recognition (ICR)
Proprietary AI trained on real Bangla handwriting: ~88% character accuracy on names, amounts, dates and form entries.
🔢
Bangla numerals preserved
০১২৩৪৫৬৭৮৯ stays Bangla in the output — dates, amounts and account numbers are never silently converted.
In the tool
📑
PDF page picker
Every page of a PDF appears as a thumbnail — pick the one to convert.
✂️
Rotate & crop
Straighten tilted photos and select only the area you need before OCR.
✏️
Editable result
Fix any word in the result box — your edits are what you copy and download.
📋
Forms as text
Filled forms come out as plain text — labels and entries in reading order, not separate fields.
Files & workflow
📄
All common formats
JPG, PNG, WebP and multi-page PDF. Phone photos supported; page previews shown beside results.
⚡
Fast processing
Printed pages in seconds; handwriting in under 60 seconds per page, with live progress feedback.
📤
Exports
Editable text, TXT, Word (.doc), JSON, HTML, printable PDF, or everything in one ZIP.
📈
Improving continuously
Verified corrections feed periodic model updates, so accuracy keeps rising on the document types you actually process.
Access & privacy
🌐
Free online tool
The online OCR is completely free with a fair daily page limit — no watermark, no payment.
🔐
Free sign-in
Your first 2 pages each day need no sign-up. For more, sign in with Google in one click or with a free email account — up to 30 pages a day and 25 MB files.
✂️
Only what you select is sent
Page choice, rotation and cropping happen in your browser; only the area you select is uploaded for OCR.
🗑️
Deletion on request
Uploads are linked to your account, so you can ask us to delete them at any time.
Ready to digitize your Bangla documents?
Upload a document and get editable text in seconds — 2 free pages a day without sign-up, no payment.