How to check spelling in text from a photo
To check spelling in text on a photo, first turn the image into editable text using OCR. You can then run a spelling check. Istio connects these two steps, so you do not have to copy the result between pages.
1. Prepare the image
You can use a photo of a printed page, a scan or a screenshot. For your first attempt, choose a clear JPG or PNG containing a single paragraph. Take the photo straight on and avoid glare. Crop out margins and neighbouring columns: extra detail can make it harder to detect the correct line order.
For a PDF, first save the relevant page as an image. For a HEIC photo, save a copy as JPG. Handwriting and complex tables are not reliable use cases for this tool.
2. Select the language and recognize the text
Open image-to-text recognition, choose a file and select the language of the text itself. An English interface does not mean the text in the image is English. Your language selection is saved in the current tab.
Start recognition. The first run downloads a language model, so it may take longer. Wait for the result; processing time varies between images.
3. Compare the result with the photo
OCR may confuse “O” with “0”, remove a line break or misread a surname. A spelling checker will not always catch these substitutions: the resulting word may also be a valid word.
Pay particular attention to dates, amounts, addresses and names. You can edit the result directly in the field below the image. If another recognition attempt fails or finds no text, the previous result stays in the editor.
4. Check spelling
Choose “Check recognized text”. Istio passes the text and selected language to the spelling page. If spelling support is unavailable for that language, the button is hidden, but you can still copy the text.
Select a highlighted passage, read the suggestion and apply a replacement if it fits. After editing, reread the entire paragraph. A message saying the service is unavailable means the check did not complete; it does not mean the text has no errors.
Example: an original typo and a recognition error
Suppose the image contains this line:
Pleese send the document by 15 November.
If OCR reproduces the line exactly, “Pleese” was already a typo in the original. The correct spelling is “Please”. If “15” becomes “I5”, that is a separate recognition error: correct it by comparing the output with the photo.
After manually comparing and editing the text:
Please send the document by 15 November.
This is a practice example, not a promise that the automated checker will return a particular suggestion. Suggestions depend on the text and the service’s rules.
Save the result
Choose “Copy” and paste the text into a document, email or note. The OCR tool does not currently offer direct DOCX or PDF downloads. A draft of the recognized text stays in the current tab so you can return to the tool. “Clear” removes that draft; it does not affect text you have copied to another editor.
Why a check can miss an error
OCR and spelling checks solve different problems. OCR reads characters, while the spelling checker evaluates the resulting text against language rules. Neither step guarantees that the meaning or facts are correct. For important documents, always compare the result with the original.
The image is recognized in your browser. When you run a spelling check, the text is sent to the Istio server.
For English sentences, see the punctuation guide for examples of direct address and comma use.
If letters are missing or lines get mixed together, follow the steps for improving recognition accuracy.