Image to Text (OCR)
Extract English text from photos, scans and PDFs with OCR, right in your browser
Your files are processed in your browser. They are never uploaded, and nothing is stored after you close the page. How we handle privacy
- Hold the phone straight above the page so lines aren't slanted.
- Use daylight or a bright lamp, with no shadow of your hand or phone.
- Scanning? 300 DPI is the sweet spot. Crop away the table or background.
- Printed text works best. Handwriting and stylish fonts often come out wrong.
Runs entirely in your browser. Nothing you type or upload is sent to us or stored.
Keep a copy of your original file. Check the result before you use it. Terms of use
How to Use
- Drop one or more photos, scans or PDFs on the box (JPG, PNG, WebP, BMP or PDF). On a phone you can take a photo of the page straight away. On a computer you can also paste a screenshot with Ctrl+V.
- Choose a Text layout: Keep line breaks for bills, addresses and lists, or Join into paragraphs for letters and book pages.
- Tap Extract text. The first time, the OCR engine (about 7 MB) downloads from this site. Each file shows its own progress in %, and every page of a PDF is read.
- Read the result in the box under each file and correct any mistakes right there. A note appears if the tool wasn't confident about the text.
- Tap Copy or Download .txt for one file, or Copy all and Download all (.txt) when you've added several.
What this image to text converter does
This image to text tool reads the printed words in a photo, a scan or a PDF and gives them back as text you can edit, copy and paste. The method is called OCR (optical character recognition). Say your college has put a circular on the notice board, or a relative has sent a WhatsApp photo of a typed letter. Instead of typing it out again, drop the picture here and you'll have the text in under a minute.
The OCR runs inside your browser using Tesseract, an open-source OCR engine. Your files aren't uploaded to us or to anyone else. That matters when the page is a bank letter, a salary slip or a medical report.
How OCR turns a picture into text
The engine first converts the image to black and white, then finds the lines of text, splits each line into words and recognises each word with a trained neural network. For every word it also reports how sure it is, as a score from 0 to 100. The tool averages those scores and shows a warning when the average is low.
below 75%: "some words may be wrong"
below 50%: "expect many mistakes"
PDF pages are first drawn as images at twice their normal size (about 144 DPI for a standard page), because small text on a page drawn at screen size is too blurry to read reliably. Small photos, such as a cropped screenshot, are also enlarged two times before reading.
A worked example
You photograph a two-page printed rent agreement with your phone. Page 1 is sharp and has 1,800 characters; the tool scores it 92%. Page 2 was taken at an angle and has 600 characters, scored 61%. The combined confidence is (92 ร 1,800 + 61 ร 600) รท 2,400 = 84.25%, so no warning appears for the file as a whole. Even so, check page 2 closely, or retake it straight on and run it again.
What reads well and what doesn't
| Source | What to expect |
|---|---|
| Scanned printed page at 300 DPI | Very good. Usually only a few small slips. |
| Screenshot of a website or PDF | Very good, as long as the text isn't tiny. |
| Phone photo of a printed page, good light | Good. Check numbers and names. |
| Photo at an angle, in dim light or with shadows | Fair to poor. Retake it if you can. |
| Handwriting, decorative fonts, text on patterned backgrounds | Poor. OCR is built for print. |
| Hindi, Gujarati or other Indian scripts | Not supported here: this tool reads English only. |
Tips for better accuracy
- Hold the phone flat above the page so the lines run straight across the photo.
- Use daylight or a bright lamp and keep your hand's shadow off the page.
- When scanning, 300 DPI is the usual sweet spot. Higher rarely helps and makes files slow to read.
- Crop out the table top, fingers and other pages before you upload.
- Always check numbers: OCR can confuse 0 and O, 1 and l, 5 and S. Recheck amounts, dates, account numbers and PAN or IFSC codes against the original.
PDFs that already have text
Many PDFs, such as e-statements and forms downloaded from government portals, already contain real text. You can select and copy it in any PDF reader, which is faster and exact. Use OCR when the PDF is a scan (you can't select the words), and use our PDF to Word converter when you want an editable document from a text PDF. Password-protected PDFs need to be opened and saved without the password first.