Read Bangla from a picture
Point it at a photo of a page, a scan or a screenshot and get the Bengali text back as Unicode — searchable, editable and ready to paste. The reader is set to Bangla before the picture arrives, so there is nothing to detect and nothing to choose. Your picture is never stored.
- Never stored
- No queue, no waiting
- No signup, no watermark
How it works
Add the picture
A photo of a book page, an old circular, a scanned certificate or a screenshot of Bangla text. JPG, PNG, WebP, HEIC and TIFF all work; a photo taken at an angle is straightened first.
Let it read
Bengali is already selected. If the page also carries English — a form, a letterhead — add it from the list and both are read.
Copy as Unicode
The text comes out in Unicode, whatever font the original was printed in: a SutonnyMJ circular reads as ordinary Bangla, not as "evsjv‡`k". Keep the layout or join the lines, then copy or download.
What Bangla does to the reading
The Bengali model reads clusters rather than letters. A conjunct such as ক্ষ or ন্ত্র is one shape on the page, and the model has to decide both what it is and how it decomposes into Unicode — which is why a misread in Bangla usually produces a wrong conjunct or a vowel sign on the wrong letter, rather than the wrong letter outright. The output is in the order Unicode expects: a vowel sign follows its consonant cluster, a reph is written as র + hasanta before the cluster it sits on, and the two-part vowels come out as one code point. That is the order every search box, spell-checker and font expects, and it is the opposite of the visual order a Bijoy typist works in.
Everything the ordinary reader does is done here too: the paper is levelled, the page is straightened, ruled tables are read cell by cell so a label and its value stay apart, and the result comes back either laid out as the page was or joined into paragraphs. Numbers are read as Bengali digits when they are printed as Bengali digits; a form that mixes ০১২ and 012 keeps both.
Bengali shares its model with no other language, but the picture may still carry English — a letterhead, a reference number, a signature block — and one language reads the other as nonsense. Adding English from the list reads the page with both packs and keeps each word in the script it was printed in.
When you want something else
A Bijoy document that still exists as a file is better converted than photographed. OCR rebuilds the text from pixels and gets a few characters in a thousand wrong on the best page; the Bijoy to Unicode converter rebuilds it from the keystrokes and gets none wrong, and keeps the bold and the tables. Photograph a page only when the file is gone.
For a whole book or an archive of scans, a desktop tool is the right shape: a command-line OCR tool with the same Bengali model reads a folder overnight, and OCRmyPDF adds a searchable text layer to every PDF in it. This page is for the page in front of you.
Frequently asked questions
Is my picture stored anywhere?
No. The language pack comes from this site the first time you use a language; the picture itself is never stored and never sent anywhere, so it works with the Wi-Fi off.
How well does it read Bangla?
Printed Bangla in a common typeface — SutonnyMJ, Nikosh, Kalpurush, SolaimanLipi, a book or a newspaper — photographed square and in focus reads well, with the odd conjunct wrong on a blurred or faded page. The mistakes are specific to the script: a reph or a vowel sign on a letter the model has not seen in that shape, and conjuncts printed small. A clear scan at 300 dpi reads better than a phone photo, and a phone photo in daylight reads better than one under a tube light.
Does it work on a Bijoy document?
Yes, and that is one of the best uses for it. OCR reads the picture of the letters, not the font's code points, so a page typed in SutonnyMJ comes out as Unicode Bangla directly — no conversion needed. For a Word file that is still in Bijoy (text, not a picture) use Bijoy to Unicode instead, which keeps the formatting and makes no reading mistakes.
Why is Bangla a separate page?
Because detection costs time and gets Bangla wrong more often than other scripts: a page with a Latin letterhead, a stamp or a number can be read as English, and the result is a confident page of nonsense. Setting the language before the picture arrives means the Bengali pack starts downloading the moment the page opens and the reader never auditions the wrong language. The ordinary Image to Text page does the same job with detection switched on, for pictures in any language.
Can it read handwriting?
Printed block letters sometimes; joined handwriting no, in Bangla or any other script — that is true of every general OCR tool. The Handwriting page says so plainly and shows what it managed.
What about a scanned PDF?
Use Bangla PDF to Word for a whole document: it reads every page the same way and gives you a .docx. This page is for one picture.
Good to know: Printed text only: joined handwriting is not recognised. Decorative Bangla typefaces, text over a photograph and very faded print come out patchy. Assamese is a separate pack — add it from the list if the page is in Assamese. The result is plain text: bold, colours and the picture itself are not carried over.
Put this tool on your website
Free for any blog, class page or help article. Paste one snippet and your visitors can use it right on your page.
Related tools
Image to Text
Words out of a picture, in 120+ languages
Bangla PDF to Word
Bengali PDFs to editable .docx, scanned or not
Bijoy to Unicode Converter
Read SutonnyMJ text as real Bangla
OCR PDF
Make a scan searchable
Screenshot OCR
Paste a screenshot, get the text
Handwriting Recognition
Block printing, read honestly