How to Copy Text from an Image or Scan (Including Hebrew)
You have a photo of a letter, a screenshot or a scanned document, and you need the text in it: to copy a number, quote a paragraph or edit it. Instead of retyping everything, you can extract the text automatically. Here's how, and what to do when Hebrew or mixed-language text comes out garbled.
Why you can't just select and copy
An image or a scan contains no text, only pixels that look like letters. The computer doesn't know there are words there. Turning them into real text takes OCR, optical character recognition, which analyzes the shapes and translates them into characters.
How to extract the text
- Open OCR.
- Upload the image (JPG or PNG) or the scanned PDF.
- Choose the language. For documents mixing Hebrew and English, pick Hebrew + English.
- Choose the output: plain text for a quick copy, Word to edit, or a searchable PDF to keep the original look.
Why Hebrew is harder
Hebrew recognition is trickier than English for a few reasons:
- Similar letters: several Hebrew letters differ only in small details, so blurry photos mix them up.
- Vowel marks: text with niqqud, as in children's books or poetry, confuses recognition.
- Mixed text: a Hebrew sentence with a phone number or an English word needs to stay in the right order.
- Stylized fonts: display fonts, handwriting-style and decorative fonts are recognized less reliably.
How to improve the result
- A sharp, well-lit photo. This matters most. Shoot straight from above, in daylight, with no shadows.
- Crop. Keep only the text. Backgrounds, margins and fingers get in the way.
- Straighten. A sideways page is recognized poorly. Rotate it first with Rotate PDF.
- Resolution. Small text needs a closer photo. If you have to zoom in to read it, OCR will struggle too.
- The right language. For pure Hebrew text, choosing Hebrew only slightly improves accuracy.
After extracting
Always review the result, especially numbers: account numbers, amounts, dates. A one-digit mistake usually doesn't stand out when reading quickly. To edit the document, choose Word. To search an archive of scans, choose searchable PDF.
What about handwriting?
Standard OCR doesn't recognize handwriting reliably, especially in Hebrew. For handwritten forms, the printed parts will be recognized well, and the handwritten parts will probably need typing.