Most of us have retyped something we already had in front of us: a society notice for the family group, a paragraph from a textbook, the address on a courier slip, a page of a grandparent’s diary. If the words are sitting in a photo on your phone, you usually don’t need to type them again. A technique called OCR can turn the picture into real, editable text that you can copy, search and paste anywhere. The quality of what you get depends heavily on the photo you give it.
What OCR actually does
OCR stands for Optical Character Recognition. To your phone, a photo of a page is just a grid of coloured dots; it has no idea that some of those dots form the word “rent”. An OCR engine looks for dark shapes on a lighter background, groups them into lines and words, and then matches each shape against the letters it has been trained on for a particular language. The output is plain text, often with a confidence score.
Two things follow from this. First, the engine has to be told which language, or script, to look for. An engine set only to English will turn a Hindi line into nonsense, because it is trying to match Devanagari shapes to Roman letters. Second, anything that blurs those shapes, such as shadow, glare, a curved page or a tilted angle, makes the matching harder and the mistakes more frequent.
How to take a photo that OCR reads well
A good photo does more for accuracy than any setting. Before you tap the shutter, run through this:
- Use soft, even light. Daylight near a window is ideal. Avoid an overhead tube light that throws your phone’s shadow across the page.
- Kill the glare. Glossy paper, laminated cards and phone screens reflect light as white patches, and text under a white patch simply disappears. Tilt the page slightly until the shine goes, and switch off the flash.
- Lay the page flat. Press a book open with your palm or a ruler at the edges so lines near the spine don’t curve away. Unfold creased letters and smooth them out.
- Shoot straight from above. Hold the phone parallel to the page, not at an angle.
- Fill the frame, then hold still. Get close enough that the text fills most of the photo, tap to focus on the words and hold steady.
- Use the full-quality image. Photos forwarded on WhatsApp are compressed and often lose fine detail. Photograph the original if you can.
Printed text versus handwriting
Be realistic here. Standard OCR engines are trained mainly on printed type: books, forms, typed letters, printed receipts. Neat, well-spaced handwriting in block letters on clean paper can sometimes work, but joined-up writing, a doctor’s prescription or a hurried shopping list will usually come out garbled. For an old handwritten letter, treat the output as a rough draft at best.
Faded thermal receipts from shops and ATMs are another hard case, because the print is grey rather than black. Raising the contrast, or turning the image into pure black and white, helps the engine separate the letters from the paper.
Hindi and other Indian scripts
Devanagari is harder for machines than English. Letters hang from the shirorekha, the line across the top of each word, and meaning depends on small marks: matras, the chandrabindu, the nukta and half-letters in conjuncts. A little blur can erase a dot or a short vowel sign and change the word. A few tips:
- Pick the right language. Choose Hindi for Hindi, Marathi for Marathi, and so on. Marathi, Nepali and Sanskrit share the Devanagari script, but each language model knows its own words better.
- Mixed pages need both languages. Bills, school circulars and forms often mix English and Hindi. Select both together so each line is read with the right model.
- Give it extra resolution. Small Hindi text in newspapers loses its matras quickly, so photograph it a little closer than you would English.
Image to Text ConverterCopy text from photos and screenshots in 22 languages, including Hindi, Bengali, Tamil and other Indian languages. Up to 25 images at once, read on your own device.
Open →
Checking and fixing the mistakes
Always read the result against the original before you use it, especially numbers. OCR mistakes look confident: the text is neat, just wrong. These are the confusions worth hunting for:
- 0 and O, 1 and l and I, 5 and S, 8 and B. These matter most in amounts, phone numbers, account numbers and PIN codes. Check every digit by eye.
- rn read as m, and cl read as d, when letters sit close together.
- Missing or wrong matras in Hindi: ि and ी swapped, a dropped anusvara, or a half-letter split into two full ones.
- Broken lines. Words hyphenated at the end of a line come out split, and page numbers or headers get mixed into the paragraph.
- Columns and tables. Two-column pages and bills can come out with lines from both columns interleaved. Crop one column at a time if that happens.
If a page comes out badly, it is usually faster to retake the photo than to correct dozens of errors.
Privacy: where does your document go?
The documents people most want to copy text from are often the most personal: Aadhaar and PAN cards, bank letters, medical reports, property papers. Many online converters upload your image to their servers to process it. You usually can’t see how long the file is kept or who can access it.
The safer route for sensitive papers is an on-device tool, where the recognition runs inside your own browser or phone and the image never leaves it. Image to Text Converter works this way: it downloads the language data once, then reads your photos locally, and you can confirm it by watching your browser’s network activity during a conversion. Either way, delete photos of ID cards from your gallery once done.
Everyday uses
- Notes and study material: pull a definition or paragraph out of a textbook photo and paste it into your notes.
- Receipts and bills: copy amounts, GST numbers and dates into an expense sheet instead of typing them.
- Notices and circulars: turn a photo of a society or school notice into text you can forward, translate or search later.
- Old letters and records: make a typed, searchable copy of printed family documents before the paper fades further.
Frequently asked
Can I copy text from a PDF this way?
If the PDF already has selectable text, just select and copy it; no OCR is needed. If it is a scan, take a screenshot of the page or export it as an image, then run that image through OCR.
Why did my Hindi come out as random English letters?
The language was set to English only. Switch to Hindi, or English plus Hindi for a mixed page, and convert again.
Do I need internet?
An on-device tool needs a connection the first time, to download the engine and the language you pick. After that, Image to Text Converter can work offline for languages you have already used.
Related guides: Compress a photo for online forms · Find and merge duplicate contacts