marathi image to text
Copy Marathi text from any photo, screenshot or scanned PDF — get editable Devanagari text in seconds. Marathi + English is pre-selected for mixed documents.
First use downloads about 8.8 MB (OCR engine + language models). After that it's cached and works offline.
About Marathi OCR
नमस्कार जग
Marathi uses Devanagari like Hindi, but with its own letters and spellings such as ळ and the eyelash reph (र्). The Marathi model knows these, so it's the better choice than Hindi for Marathi newspapers, books, 7/12 extracts and notices from Maharashtra.
Because Marathi documents often include English terms and numbers, Marathi + English is selected by default. The output is Unicode Marathi that you can paste into any app.
Tips for accurate Marathi text
- Use the Marathi model, not Hindi, for Marathi text — it handles ळ and Marathi spellings properly.
- Photocopies of photocopies lose detail; use the cleanest copy you have.
- If a page has both Marathi and Hindi, select both languages.
- Crop to the text you need and avoid shadows across the page.
How it works
The tool runs Tesseract's neural-network (LSTM) Marathi model directly in your browser. Small images are enlarged first, because OCR works best when letters are 20–40 pixels tall. The first time you use it, your browser downloads the OCR engine (about 4 MB) and the Marathi model (about 1.9 MB); after that, everything loads from cache. You get plain Unicode text that you can edit, copy, or download as a .txt file.
Other languages
- English OCR
- Hindi OCR
- Tamil OCR
- Telugu OCR
- Bengali OCR
- Gujarati OCR
- Kannada OCR
- Malayalam OCR
- Punjabi (Gurmukhi) OCR
- Urdu OCR
- Arabic OCR
- French OCR
- Spanish OCR
- German OCR
Frequently asked questions
Is Marathi OCR different from Hindi OCR?
Both use Devanagari, but the Marathi model is trained on Marathi words and letters, so it's more accurate for Marathi text.
Is Marathi OCR free?
Yes. There's no sign-up or page limit. The Marathi model (about 1.9 MB) downloads once and is cached by your browser.
Are my images uploaded to a server?
No. Recognition runs on your device using WebAssembly. Your images and PDFs never leave your browser.
Can I convert a Marathi PDF to text?
Yes. Drop the PDF — each page is rendered and read in your browser, up to 30 pages at a time.