How to translate a PDF between Bengali and English
Updated 19 September 2026
Translating a PDF is two problems wearing one coat: getting the words out of the file, and translating them. Bengali makes the first problem harder than English does, and that is where most attempts fail.
Check whether the file has real text
Try to select a line. If the characters highlight, the text is really there and translation will work directly. If nothing selects, the page is an image and no translator can read it yet.
There is a third case peculiar to Bengali and other Indic scripts: text that selects but comes out as nonsense when copied. That happens when the PDF was made with a legacy font that stores its own character codes rather than Unicode — common in older Bangladeshi government documents and anything made with Bijoy. It looks like text and behaves like a picture.
A scan needs nothing from you
Translate PDF checks whether the page has any text in it. If it does not, it reads the page with OCR itself, as part of the same run, and translates what it reads — you do not have to run anything first.
The legacy-font case is the one worth knowing about. Text that selects but copies as nonsense was written with a font that stores its own character codes instead of Unicode — Bijoy, and much older Bangladeshi government paperwork. To a translator it looks like text and behaves like a picture, so put it through OCR yourself: OCR reads the shapes on the page rather than the codes behind them.
Then translate
- Open Translate PDF and choose your document.
- Pick the language you want it in — English is the default, and Bengali is in the list.
- Choose faithful for official documents, plain language for something you only need to understand.
- Download it as a PDF with the original design, or as a Word file if you want to edit the translation.
What you get, and what you do not
- You get the document back looking like itself: the same fonts, sizes, colours, headings, tables and page breaks, with Bengali set in a font that has Bengali letters. A sentence that grows in the new language is fitted to the room it has rather than printed over the line beneath it.
- A scanned page comes back as the scan, with the translation set over it — the letterhead, the stamp and the signature stay where they were.
- You do not get a guarantee about names, addresses and numbers. Check those by eye: machine translation transliterates Bengali names inconsistently.
Official documents
A machine translation is not a certified translation. Embassies, courts and universities generally require a translation signed by a recognised translator. Use this to understand a document, to check a translator’s work, or for anything unofficial — not as the certified copy itself.
Questions
- Why does copied Bengali text come out as gibberish?
- The PDF uses a legacy non-Unicode font, such as Bijoy, which stores its own character codes. Your computer maps them to the wrong letters. Running the file through OCR produces real Unicode text.
- Is translating a PDF here free?
- There are free AI runs every day, and a long document uses one for every few pages. Beyond those, runs are charged by what they actually use — a few cents for a long document — from a balance you top up.
- Can I translate into Hindi, Urdu or Arabic too?
- Yes. Twenty languages are in the list, and right-to-left scripts like Urdu and Arabic render correctly in the result and in the Word download.