Whole Urdu and Arabic books to editable Word, checked word by word.
Upload a complete book, up to 1 GB. Several AI readers read every line, a dictionary checks every word, and you fix only the few we flag. Every page comes back in its own layout.
- العربية
- اردو
- فارسی
- हिन्दी
- English
- Français
- Deutsch
- Español
- Italiano
- Português
- Türkçe
Drop a scan here, or choose a file
PDF, JPG, PNG, TIFF, DOCX or XLSX, up to 1 GB (whole books welcome)
Free quality check first. Nothing is OCR'd until you confirm.
Private to your account and encrypted. Delete it any time; files are removed after 30 days. Privacy
What comes back
From a printed page to text you can edit
Nastaliq and Naskh, columns and headings, read in the right order. The one word the readers disagreed on is waiting for you, with their readings one keystroke away.
Your scan
Your text
کتاب انسان کی بہترین دوست ہے۔
پرانی کتابوں کے صفحات میں ہماری تہذیب
کی کہانی محفوظ ہے، مگر کاغذ وقت کے ساتھ
کمزور ہو جاتا ہے۔
Everything you need after the scan
A Word file that looks like the original
Headings, paragraphs, tables and verse, in the right direction, ready to edit and print.
A review edition, page beside text
Each scanned page next to its text, with uncertain words highlighted, for proofreaders and editors.
Listen to it
Hear any page read aloud in a calm news-anchor voice while you check it, or make an audiobook of the whole document.
Send to LexaTranslate in one click
The verified text goes straight to LexaTranslate, with no copying and pasting.
For libraries and publishers
Digitise a whole collection, not one page at a time
LexaOCR is built for complete books and archives: back catalogues, newspaper runs and rare printed books. Upload whole volumes, let the agents read them on our servers, and give your editors only the words that need a human eye.
- Whole books in one upload, up to 1 GB each, read page by page on our servers
- Word editions that keep the layout, plus a review edition for your proofreaders
- Audiobooks of your back catalogue, and one-click translation with LexaTranslate
- Volume pricing, invoices in AED, a data-processing agreement and priority support
Meet the team that reads your pages
Eleven agents work on every document. Several readers read each line independently, Mizan holds the vote, Adib checks the dictionary, Daleel shows you only the words that still need a human eye, and Naqsh gives you the text back in the document's own layout: tables, forms, passports, articles and poetry.
- NoorInspector
- MistarLayout
- SafaRestorer
- RaqimPage reader
- BasiraVision reader
- KatibLine reader
- KhattatNastaliq reader
- MizanVote
- AdibLinguist
- DaleelRouter
- NaqshFormatter
يقرأالطلابفيالقاعةالكبيرةكلصباح
What happens to your document
- 1
Quality check before you start
Digital PDFs are extracted instantly. Scans get a quality score and an accuracy forecast per page, so you can rescan blurry pages first.
- 2
Restored, never blurred
Pages are straightened and cleaned without smoothing filters, so the dots that separate ب ت ث ن ي and accents like é è ê survive.
- 3
Read by several readers at once
Independent readers, including one for Urdu Nastaliq, read every line in parallel. Where they agree, the word is almost certainly right.
- 4
Checked by a linguist agent
Disagreements go to a dictionary for the page's language (3.1 million Arabic words, 146,000 Urdu words, over 500,000 each for French, German, Spanish and Portuguese), and to a linguist AI when the dictionary can't decide.
- 5
You review only the red words
Everything verified stays out of your way. You see the handful of uncertain words, with the readers' versions one keystroke away.
- 6
Back in its own format
Passports, ID cards, forms, invoices, tables, letters, newspapers and poetry come back laid out like the original: fields, real tables, headings, columns in reading order.
Questions
Is my document private?
Yes. A document is visible only to the account that uploaded it. Files are encrypted in transit and at rest, and our staff do not open them unless you ask us to for support, or security or the law requires it.
How long are my files kept?
Files are deleted automatically 30 days after upload, and you can delete a document yourself at any time.
How well does it read Urdu Nastaliq?
LexaOCR has a reader made for Nastaliq and checks Urdu against a 146,000-word dictionary. Results depend on the print and the scan, so the free quality check forecasts accuracy for every page before anything is read.
Which languages are supported?
Urdu, Arabic, Persian, Hindi, English, French, German, Spanish, Italian, Portuguese and Turkish, including pages that mix them.
Can it read handwriting?
Not yet. LexaOCR is built for printed text: books, newspapers, magazines, forms, IDs and typed letters.
What do I get back?
Editable text in the original layout, a clean Word file, a review edition with each page beside its text, a searchable PDF and plain text. You can also listen to the pages and send the text to LexaTranslate.
What does it cost?
LexaOCR is free during the beta. The plans that follow are on the pricing page.
Do I need an account?
Yes, a free Lexa account. If you already use LexaTranslate, the same account works here.
From scan to translation in one family
Send the verified text straight to LexaTranslate, where 15 cultural translation agents take it into English and 14 other languages.
Try it on one page
Upload a single page first. The quality check is free, and nothing is read until you confirm.