The broken-Arabic problem
If you've ever copied Arabic out of an old PDF and got disconnected letters running in the wrong direction, you've hit the classic issue: the file stores visual glyphs, not real text. OCR fixes that by reading the image and writing proper Unicode Arabic in reading order, whatever produced the original.
What it's used for
- Copying passages from Arabic books and textbooks
- Extracting text from screenshots of Arabic posts, articles and slides
- Digitising printed letters, contracts and official documents
- Reading signs, menus and product labels
- Pages that mix Arabic and English, with each kept in its own direction
Persian, Urdu and other Arabic-script languages
The Arabic page tells the model to expect Arabic, but the script is shared with Persian, Urdu, Pashto and Kurdish (Sorani). Extra letters such as پ چ ژ گ are read correctly. For mixed or unknown languages, the general handwriting OCR detects the language automatically.
Images only
This tool accepts JPG, PNG, WebP and HEIC images. For a PDF, screenshot or export each page and upload up to five at once.
Mostly handwritten pages? Try Arabic handwriting to text.
Frequently asked questions
Is this Arabic OCR free?
Your first image is free. A free account gives 3 credits and then 1 free credit per day.
Can I convert an Arabic PDF to Word with this?
Screenshot the PDF pages, run them through this OCR, and paste the text into Word. Direct PDF upload isn't supported here.
Does it keep harakat (tashkeel)?
Yes, if they appear in the image. It doesn't add diacritics that aren't there.
Does it work on Quranic or classical texts?
It reads clearly printed classical text well. For religious texts, always verify against an authoritative printed edition.
Which image types are supported?
JPG, PNG, WebP and HEIC, up to 10 MB each.