Image to Text (OCR) — free, private, in your browser
Image to Text (OCR) reads the text in one image or a batch of up to 20 using the Tesseract.js OCR engine, which runs inside your browser, in any of 13 languages, and gives you each result in an editable box. Free, no signup, 5 runs a day. It runs in your browser; nothing you add is uploaded.
🔍 Image to Text (OCR)
Read printed text in 13 languages from up to 20 images at once
Extract text from images using AI-powered OCR
mdash; in your browserA batch of up to 20 images counts as one of your 5 runs today. The first run in a language downloads that language's data (about 1–20 MB); your images stay on your device.
About Image to Text (OCR)
Pull editable, copyable text out of a photo, screenshot, or scan using on-device OCR. TaimurTools runs the recognition entirely in your browser, so the image is never uploaded — ideal for extracting text from documents, whiteboards, or receipts without handing them to a cloud service. No signup, no watermark. Just drop in an image and copy the recognized text.
- On-device OCR — image never uploaded
- Copy or download the extracted text
- No signup or watermark
- Free, no signup
Step-by-step guide: How to convert an image to text free (OCR)
Runs on your device
Your file is opened and processed by JavaScript inside this page. It is never uploaded and we never receive it. The page does download the processing library it needs from a public CDN the first time you use it - that request carries no part of your file.
Free tier: 5 runs per day for this tool, counted in your browser. The counter resets at midnight UTC and is not shared between browsers or devices.
How TaimurTools handles data · Why most of these tools run on your device
What this tool is for
- Getting text out of screenshots, several at a time
- Copying a printed notice in Urdu, Arabic, Hindi or another supported language
- Pulling a serial number or an address off a picture and fixing any misread character before copying
| What it does | Reads the text in one image or a batch of up to 20 using the Tesseract.js OCR engine, which runs inside your browser, in any of 13 languages, and gives you each result in an editable box. |
|---|---|
| Accepted input | Up to 20 images per run (PNG, JPG, WebP, GIF or BMP), plus the language of the text: English, Urdu, Arabic, Hindi, Spanish, French, German, Italian, Portuguese, Turkish, Russian, Chinese (Simplified) and Japanese. For a non-English language you can also have it read English words mixed into the text. |
| What you get back | One editable text box per image. Copy one, copy all, or download everything as a single .txt file (with the file name above each image's text). Your edits are what gets copied and downloaded. |
| Limits | Capped at 5 runs per day for this tool. A batch of up to 20 images counts as one run, and a run only counts when some text is found. Printed text only - not handwriting. |
| Browser requirements | A current browser. The OCR engine downloads from a CDN on first use, and each language's data downloads the first time you pick it (roughly 1-20 MB). None of that carries your images. |
A worked example
- Input
- Three screenshots of an invoice, a blank image and a meeting slide, language English.
- What comes back
- Three boxes: "Invoice number 4471", an empty box marked "No text detected", and "Meeting at 3pm Friday" - one run used.
- Worth knowing
- Accuracy tracks image quality. Sharp, high-contrast printed text reads well; small, blurry, angled or low-contrast text and handwriting produce errors. Each of the 13 languages read a clean rendered test line perfectly in our checks, but real photos are harder - proofread codes, amounts and dates.
If it does not work
"No text detected"
The text is too small, too faint or at too much of an angle, or the language setting does not match the text. Crop tighter, use a larger, straighter image, and check the language.
The text is garbled or in the wrong alphabet
The language setting probably does not match the text. Pick the right language and run it again.
Chinese or Japanese has odd spacing
The tool joins characters back together automatically; if a space remains, delete it in the box before copying.
Handwriting was not read
Tesseract targets printed text; handwriting is out of scope.
Some files were skipped
Only images are accepted, and at most 20 per run. Convert HEIC photos first with Image Converter.
Nothing happens on Extract Text
You may have used your 5 runs for this tool today.
How to use Image to Text (OCR)
- Pick the languageChoose the language of the text. For Urdu, Arabic or another non-English language, leave "Also read English words" ticked if the text mixes in English.
- Add imagesDrop up to 20 images or click to choose them. Remove any you added by mistake with the x button.
- ExtractClick Extract Text. Each image is read in turn; the first run in a language downloads that language's data first.
- Fix, then copy or downloadCorrect any misread words in the boxes, then Copy one result, Copy all, or Download .txt.
When to use Image to Text (OCR)
Getting text out of screenshots
An error message, a table in a screenshot, or a chat that was sent as a picture can be turned back into text you can search for or paste - several screenshots in one go.
Reading Urdu, Arabic or Hindi text
A photographed notice, a poster or a page in Urdu, Arabic, Hindi or one of the other supported languages becomes text you can copy, with English words mixed in still read correctly.
Copying a printed notice or a photographed slide
A photo of a notice board or a presentation slide gives you editable text instead of retyping it.
Pulling a serial number or address off a picture
OCR gives you the characters to paste. Because the result is editable, you can fix a misread 0 or O before you copy it.
Digitising several printed pages
Photograph up to 20 pages and read them in one run; Download .txt gives you all of them in one file, each under its file name.
The settings, explained
Pick a language, add up to 20 images and click Extract Text. The Tesseract.js recognition engine runs inside your browser:
- Language of the text
- English, Urdu, Arabic, Hindi, Spanish, French, German, Italian, Portuguese, Turkish, Russian, Chinese (Simplified) and Japanese. Each language was checked end to end against a rendered test line before it was listed. Pick the language the text is written in - the wrong one produces garbled output.
- Also read English words
- Shown for every language except English. It loads English alongside, so Urdu or Arabic text with English brand names or numbers reads correctly. Untick it for text with no English in it.
- Batch of images
- Up to 20 images per run, read one after another. The whole batch counts as one of your 5 daily runs.
- Editable results
- Each image gets its own text box next to a thumbnail. Fix misread characters there - Copy, Copy all and Download .txt use your edited text.
- First run in a language
- The engine and that language's data download from a CDN the first time, so the first read takes longer. That download is the engine, not your images.
- When a run counts
- A run only counts when at least one image produced text, so a batch of blank images does not use one of your 5 runs.
What it cannot do
- Accuracy follows image quality. Sharp, high-contrast printed text reads well; small, blurry, angled or low-contrast text produces errors, and handwriting is not supported.
- The result is plain text. Layout, columns, tables and fonts are not reproduced, and a table can come out as lines you will have to rebuild.
- One language setting applies to the whole batch. Split images in different languages into separate runs.
- Always proofread anything that matters. OCR confuses characters that look alike: the letter O and the digit 0, I and l and 1, rn and m. Check codes, amounts and dates against the original.
How it compares with the alternatives
Phones and some operating systems can already select text inside photos (Live Text on iPhone, the text selection in Google Lens), and those handle many languages. If you are on a phone and the text is short, use that.
Image to Text suits a desktop browser, a screenshot that lives on your computer, or a picture you would rather not hand to a cloud service: recognition happens in your browser tab, so the image is not uploaded to this site or to anyone else.
How in-browser processing works, and exactly which tools are the exception, is explained on the why client-side page.
Getting a better result
Match the language
The language setting matters more than anything else. Urdu text read as English comes out as nonsense.
Crop to the text
Cut away borders, logos and photos around the text. Less clutter means fewer false characters.
Use the largest, sharpest version
A screenshot taken at higher zoom reads better than the same screenshot shrunk down.
Fix in the box, then copy
Correct numbers and names in the editable result before you copy it anywhere.
More questions about Image to Text (OCR)
Why did it return nothing?
The picture may have too little contrast, the text may be too small, the language setting may not match, or it may be handwriting. A run with no text found is not counted.
Which languages does it support?
English, Urdu, Arabic, Hindi, Spanish, French, German, Italian, Portuguese, Turkish, Russian, Chinese (Simplified) and Japanese.
How many images can I do at once?
Up to 20 per run, and the batch counts as one run.
Does it keep the formatting?
No, only the words, in reading order.
Does the first run take longer?
Yes. The engine and each language's data download once; later runs are faster.
Related tools
- OCR PDF - do the same for multi-page scans, in 12 languages
- Image to PDF - wrap pictures in a PDF to use OCR PDF
- Image Enhancer - sharpen a soft photo before reading it
- Image Resizer - enlarge or shrink the picture first
Step-by-step walkthrough: read the full guide.
We automate document parsing, summaries, data extraction and lead enrichment inside the tools you already use, so the work runs without you. The free tool above stays free.
Not sure what to automate first? Take the free AI Automation Opportunity Audit →Frequently asked questions
Is my image uploaded to a server?
No. The Tesseract.js engine runs inside your browser, so the images never leave your device. The engine and language data are downloaded to you from a CDN - that carries no part of your images.
Which languages can it read?
English, Urdu, Arabic, Hindi, Spanish, French, German, Italian, Portuguese, Turkish, Russian, Chinese (Simplified) and Japanese. Pick the language before you extract; mixed Urdu or Arabic text with English words reads best with "Also read English words" ticked.
Can I process several images at once?
Yes - up to 20 per run, and the whole batch counts as one of your 5 daily runs. Each image gets its own editable result.
Can I correct mistakes before copying?
Yes. Every result is an editable box, and Copy and Download use your corrected text.
Does it read handwriting?
No - printed text only.
Is there a limit?
Yes - 5 runs per day for this tool, a batch counts as one, and a run with no text found is not counted.