Image to Text Converter (OCR)
Choose an image, press Extract text, and copy or download the words found in it - the picture never leaves your device.
Text is read inside this page by an engine that runs in your browser. Your images are never uploaded.
What this tool does
To get text out of an image, choose the picture, pick the language and press Extract text. An OCR engine running in your browser reads the characters and gives you editable text to copy or download. Because it runs locally the image is not uploaded. It works best on clear, printed text and is poor at handwriting.
Runs the open-source Tesseract OCR engine inside your browser on each image and returns the text it recognises.
Enlarges small images first, can join broken lines into paragraphs, and reports how confident the engine was.
How to use it
- Choose one or more images, or drop them onto the box.
- Pick the language the text is written in.
- Leave "Enlarge small images" on for screenshots and phone photos.
- Press Extract text; the first run downloads the engine from this site.
- Proofread the result, then press Copy text or Download .txt.
Why use this converter
Reads the text out of an image with an OCR engine that runs in your browser, so the picture is never uploaded. Text trapped in a screenshot, a photo of a page or a scanned form cannot be searched, edited or quoted. OCR turns it back into real text in seconds, and doing it in the browser means a private document never has to be sent to a stranger's server.
Why we built it
We built the image to text converter to run the OCR engine in your browser, so the picture never leaves your device. Read scanned documents with the OCR PDF converter and count the result with the word counter.
If your task goes further, try the Image to ASCII Art Converter.
Who this tool is for
- Students photographing notes
- Office staff digitising printed pages
- Anyone copying text from a screenshot
When to use it
- Copying text from a screenshot, a slide photo or an image in a chat.
- Digitising a printed page, receipt, letter or book excerpt.
- Pulling a quote or a figure out of a scanned document.
- Making an image's text searchable, or reading it with a screen reader or text-to-speech.
How it works
The image is first drawn onto a canvas at a size that suits recognition: very large pictures are reduced, and small ones are enlarged up to three times, because the engine reads best when the letters are about 20 to 30 pixels tall. Transparent areas are filled with white.
The Tesseract engine then finds lines and words, recognises each character with a neural network and returns the text and a confidence score. Everything happens in a background worker in your browser using WebAssembly, with the engine and the language data served from this site.
OCR is good at clear, printed, horizontal text on a plain background and poor at handwriting, stylised fonts, curved or rotated text and low-resolution photos. Always proofread - numbers and similar-looking characters such as 0/O and 1/l are the usual mistakes.
What you give it, and what you get back
| Item | Details |
|---|---|
| Images | PNG, JPG, WebP, BMP or GIF files, up to 20 at a time. Larger than about 3,000 pixels on the long side is reduced. |
| Language | The language the text is written in. Recognition is much worse if it does not match. |
| Options | Enlarge small images, and keep line breaks or join lines into paragraphs. |
| Output | Editable text in the box (with a separator per file if you chose several), word count and the engine's confidence, plus copy and .txt download. |
Issues you may run into
| What you see | Why it happens | What to do |
|---|---|---|
| The text is full of mistakes | The image is blurry, small, skewed or low-contrast, or the language is wrong. | Use a sharper, straighter, higher-resolution image with dark text on a light background, and check the language. |
| Nothing is found | There is no machine-readable text, or it is handwritten or very decorative. | Try a cleaner crop of just the text, or retype it. |
| It says the engine could not start | The browser blocked WebAssembly or workers, or a network error interrupted the first download. | Use a current browser, allow the site to run scripts, and try again. |
| Columns or tables come out in the wrong order | The engine reads in lines across the page. | Crop each column or cell separately, or reorder the text afterwards. |
Assumptions and limitations
- Works best on printed text; handwriting is not reliably recognised.
- Seven languages are included; scripts such as Arabic, Chinese, Japanese and Cyrillic are not.
- No layout is kept - only text, in reading order - and tables and columns are not reconstructed.
- The first run downloads a few megabytes of engine data from this site, then the browser keeps it.
Frequently asked questions
How do I extract text from an image?
Choose the image, select its language and press Extract text. The words appear in the box on the right, ready to copy or download.
Is my image uploaded to a server?
No. The OCR engine runs inside your browser and the image never leaves your device. The engine and language data are downloaded from this site once, not from a third party.
Why is the text inaccurate?
The usual causes are a low-resolution or blurry image, poor contrast, a tilted photo or the wrong language. Clear printed text on a plain background gives the best result.
Can it read handwriting?
Not reliably. The engine is designed for printed text. Neat block capitals sometimes work; cursive handwriting almost never does.
Which languages are supported?
English, Spanish, French, German, Portuguese, Italian and Hindi. Choose the language of the text before you start.
Can I extract text from a PDF?
For a PDF with real text, use the PDF to text tool. For a scanned PDF, convert its pages to images with the PDF to image tool and run them through this one.