How to Extract Editable English or Urdu Text From an Image
Turn printed words in a screenshot or photo into editable text, then check the output before copying it.

Turn a photo into editable text
Open the Image to Text tool and choose a JPG, PNG or WebP image. Select English, Urdu or English + Urdu, then run the extraction. Look at the output alongside the source. Edit mistakes and copy or download the text only after you are satisfied with the words, numbers and punctuation.
The process is called optical character recognition, or OCR. The tool looks for readable text in an image and produces characters you can work with in a document. A screenshot of a table may yield a plain-text version of the contents rather than an editable copy of the entire layout. For a form or printout, keep the source if the placement of information matters.
FixHubLab downloads its OCR engine and selected language data when you start extraction. The image is processed on your device. The first run may take longer because the language files have to load. A confidence score can give you a rough indication of recognition quality, but check anything that matters against the actual picture.
Decide what you need from the image
If you need a quotation from a printed page, you may only need one paragraph. Crop to that paragraph while leaving all of its lines visible. If you need a table of figures, plan to check the extracted data row by row. If you are gathering notes in two languages, choose a workflow that preserves which words belong to which sentence.
A different job may need the original visual appearance. A signed letter, a receipt with a visible stamp, or a screenshot that shows where a button is located can be more useful as an image. OCR can provide an editable transcription for search or copying, while the original can show what the words looked like and where they appeared.
Make the distinction before you spend time cleaning the result. A copied phone number may need only an absolute accuracy check. A full page for a newsletter may need reformatting, supply of missing headings and review by a person familiar with the language. Keep both artifacts if they serve different purposes.
Take a photo that is easy to read
Place the document on a flat surface in even light. Hold the camera parallel to the page and leave a little room at each corner. Look closely at the preview before moving away. Check whether the smallest text is in focus and whether a hand, dark shadow or reflection crosses a line you need.
A tilted photograph can make individual lines harder to recognise, especially when they become small toward one edge. Retake it from above when possible. If you cannot retake it, crop away the background and correct the rotation in the Image Editor. Compare the prepared copy with the original to make sure the crop did not remove punctuation or page labels.
For a glossy sheet, move the light or camera slightly until the glare disappears. A bright reflection can hide characters completely. Increasing contrast after the photograph is taken cannot recover words lost to the reflection. A second photograph with good lighting often saves more time than trying several processing settings on an unreadable source.
Choose English, Urdu or both
Use English for a page that is entirely English. Use Urdu for a page that is entirely Urdu. Choose English + Urdu when both languages appear in the same image and you need text from both. For example, a notice might have an Urdu instruction on each line and a short English label next to it. Choose both if those labels are important.
If the languages appear in separate blocks, you can also crop them and run each block in its own language. This may make the output easier to organise, especially when one half of the image is a translation of the other. Keep a record of which block came first and which heading belongs to each part.
Urdu script often presents additional layout challenges when combined with numbers, English abbreviations or web addresses. Read the extracted text in an editor that handles right-to-left paragraphs. A line can appear to move punctuation or numbers when displayed even though the underlying characters are present. Compare it with the original and adjust the paragraph direction in the destination editor when appropriate.
Worked example: copy English text from a screenshot
Imagine a screenshot contains a headline, two paragraphs and unrelated navigation along the top. Crop away the navigation, keeping the headline and both complete paragraphs. Choose English in the Image to Text tool and start extraction. Review whether the headline and first paragraph are in the right order.
Look at small punctuation such as an apostrophe, quotation marks and a period. Check any name that uses unusual spelling. If a word wraps at the end of the screenshot line, decide whether the break is part of the word or only the way it was displayed. Rejoin the sentence when copying it into a document that wraps words differently.
Compare the cleaned excerpt with the original once more. When only one paragraph matters, there is little value in correcting text from parts you will not use. Focus your review on what you will publish, search, quote or send. Retain the source screenshot so you can refer to it if a wording question comes up later.
Worked example: an Urdu notice with English dates and labels
Suppose an Urdu notice includes English office names and dates written in numerals. Start with a clear image in the correct orientation and select English + Urdu. After recognition, read the Urdu text from the source and compare each line with the output. Pay extra attention to small marks and letters that are close together in the print.
Copy the output into an editor that supports right-to-left paragraphs. Check dates and English labels individually. For example, a day and month can appear in an unexpected order depending on how the software follows the direction of the surrounding text. Keep the source beside the editor so you can compare the date with the way it is printed.
If the extracted text from a bilingual line looks scrambled, try processing its parts separately. Crop one block into a working image, select the language that applies and compare it with the previous result. Reassemble the text yourself if the order is important. The aim is an accurate, readable transcript, even if it takes more than one pass.
Process columns, tables and small print separately
OCR may read across two columns on the same line, or interleave headings with paragraphs. If the result is hard to follow, crop the first column and extract it on its own. Then crop the second and repeat. Join the text only after confirming which column comes first. Leave enough margin to include letters at the edge of each block.
A table needs its own check. Make a new table in the destination application, then compare each extracted entry with its row and column. A misplaced value can change the meaning even if OCR recognises all the digits correctly. Check empty cells and repeat marks carefully. For a printed address list or product catalogue, label each group as you work through it.
Small print may need a closer source photograph rather than a larger display of the same image. If you have the original document, capture the difficult section separately. Make sure the marks on letters and the difference between a decimal point and a speck are visible. Then run OCR on that crop and use the result to correct the working transcript.
Check names, amounts and punctuation first
Some characters look similar. A capital O may look like zero, and a lowercase l may look like the number 1. In a serial number or invoice reference, those substitutions can make the entire value unusable. Compare each character with the source rather than relying on the visual similarity of the output.
Check decimal points, commas, minus signs, slashes and percent signs. A missed punctuation mark in ordinary prose may be easy to fix later, but it can change a date, quantity or amount. Zoom in on the source to decide whether a dot is a real decimal point or a mark on the paper. If the text will drive a payment or decision, verify it against the original record.
For names, respect the spelling in the source. Do not silently normalise a name because a version you expected is more familiar. If the scan is unclear, mark the uncertain portion in your working copy and ask an appropriate source to confirm it. A clearly marked uncertainty is more useful than a confident but incorrect transcription.
Copy a clean result into the destination
After checking the text in the browser, copy it into the document or notes app where you need it. Inspect line breaks, spaces and paragraph direction. Browser OCR may retain a line break from the source because it was part of the visual layout. Remove breaks inside sentences when a normal flowing paragraph is the intended result.
A text file is useful for storing or reviewing a transcription, especially when you need to pass the text to another program. Plain text will not keep the source’s fonts, columns or boxes. Add deliberate headings and lists in the destination if they help the reader. Do not assume a plain text download reproduces a table as cells.
If you need a visual document to share alongside the text, use the original image or put a set of images into a PDF. Our guide to turning images into one PDF explains how to create the visual copy. Confirm which artifact the recipient requested so you do not send a transcript when they need the original layout.
What to do when recognition looks wrong
First, try the language that matches the printed text. If the page is mostly one language with a small block in another, test both mixed recognition and a separate crop of the difficult block. If lines are being joined incorrectly, remove nearby images, borders or other columns from the working copy.
Try a clearer input if the output is empty or full of errors. Check that the photo is facing upright and reading order runs naturally across the image. Crop to one or two lines and test those first. Once a sample works, expand the selection or prepare the rest of the page the same way. This is often quicker than repeatedly processing the full document.
Handwriting, heavily stylised text, unusual fonts and faint images can be harder than clear printed text. If the source cannot be improved, manually transcribe the important portion and mark what you cannot read. An OCR result is a starting point for review. It should not replace that review when the document must be accurate.
Handle the image, download and browser limits
The tool accepts JPG, PNG and WebP up to 20 MB and 12 million source pixels. File size and pixel count are measured separately. A clear photograph can be small in megabytes and still have more pixels than the tool accepts. If necessary, crop to the relevant text or resize a working copy. Avoid making the letters too small to read.
If the first OCR download fails, check the network connection and try again. The first use of a language can require provider files to load. The image itself is processed on your device, but the download providers receive ordinary request information. For details, see the FixHubLab Privacy Policy. Keep the source private and follow any rules that apply to it.
Use caution when shrinking a detailed image to fit a limit. A crop of the required text can preserve more useful detail than reducing a whole page. If a compression step makes small marks fuzzy, return to the original and prepare a clearer working copy. Our image compression guide covers how to reduce size while checking visible details.
Review a batch before transcribing every page
Start with one simple image and one difficult image from the batch. Choose the language and try the same preparation workflow on each. If the simple page is accurate but the difficult one has multiple columns, handle the columns separately. If both are inaccurate, recheck the source orientation, lighting and text size before processing more pages.
Keep the image filenames or page positions next to each excerpt while you work. This helps if a number needs to be checked later. You might use a separate note for page 1, page 2 and page 3, or mark source references next to individual values. Organise them before combining the text so it remains clear what came from each image.
When a second person will review the transcription, share the relevant source and point out anything you marked as uncertain. A short list of names, dates and amounts to check can make their review more effective. Do not assume they will know the original layout merely from receiving a page of plain text.
Preserve enough context to verify a quotation
If you are copying a passage for a reference, record the source title and page position before you crop. A tightly cropped photograph may omit the name of the author or the page number you need later. Keep a wider source image and a separate crop for recognition. That way you can use a clean crop without losing track of where it came from.
Compare the passage with the source immediately before quoting it. Look for words at the end of a line, where a tight crop can hide the last letter. Check punctuation around quoted speech and any brackets, footnote marks or unusual names. A word that is repeated in the original may be intentional. Fix transcription errors without silently changing the wording you mean to quote.
Sometimes the source itself is unclear. For example, an older photocopy may have a dark mark over a date or a tear through a letter. Try a second photograph or another copy of the document before guessing. If the uncertainty matters and another source is unavailable, keep a note of the unclear portion. That record makes it easier to revisit the text later without treating an ambiguous reading as settled.
When combining several excerpts, use a source label above each one. It can be as simple as the filename and page position. This helps distinguish words that were actually adjacent in the original from passages gathered from several places.
A final checklist for a usable transcription
Check the language, source orientation and legibility before recognition. Crop away distractions where practical. After extraction, compare the text with the image and inspect names, punctuation, dates and figures. Use a destination editor that displays right-to-left text correctly if working with Urdu.
Keep the original image available in case you need to verify the result. Copy or download the clean transcript when it is ready. Where the source layout matters, send a visual copy separately. When the task only requires a few words or a number, a small careful crop and a final comparison may be all you need.
