Free Image to Text
Extract text from images locally with English and Spanish OCR. Preview crop, rotation and contrast, process a batch, review confidence details and download editable text or word data.
Loading tool…
How to extract editable text from images in your browser
Read English or Spanish text from images with local OCR. Preview crop and contrast adjustments, extract a batch and review editable text and confidence details before downloading.
- Choose, drop or paste up to ten supported images, or add an English or Spanish sample.
- Select the recognition language and page layout. Adjust crop, rotation, scale or contrast and preview the prepared image.
- Choose Extract text and wait for the local OCR model and batch processing. Review low-confidence words against the images.
- Edit the combined text and download TXT or JSON. Use the per-image bundle for individual text and TSV word data.
Who can use image OCR
Office staff can pull text from a screenshot of a table or a photo of a printed letter instead of retyping it. Students can capture quotes from a book page they photographed in the library.
Small business owners can extract totals and dates from receipt photos, and bilingual teams can read English and Spanish documents in one pass. Because the images stay on your device, it also suits documents you would rather not upload anywhere.
A worked example with crop and scale
Suppose you have a 1500 x 1000 pixel phone photo of a receipt, and the total sits in the middle. Open the crop settings and enter left 10, top 20, width 50 and height 30. The crop starts 150 pixels from the left and 200 pixels from the top, and it is 750 x 300 pixels.
Set scale to 2x, and the prepared image becomes 1500 x 600 pixels, or 0.9 megapixels, well under the 16 megapixel limit for prepared images. Choose Single text block for the layout, turn on Grayscale, and use the Digits only preset if you only want numbers.
After extraction, set Review words below confidence to 80 percent. Any word the engine scored below 80 appears as a chip with its score, for example a smudged 7 that might really be a 1. Compare those words with the photo before you trust the total.
How recognition and confidence work
The tool uses the open-source Tesseract OCR engine running in your browser, with English, Spanish or combined language models. The page layout choice tells the engine what to expect: Automatic page, Single text block, Scattered text, Single line or Single word. Picking the right one often improves results more than any other setting.
Preprocessing runs before recognition. You can rotate by 90, 180 or 270 degrees, scale from 0.5x to 3x, convert to grayscale, invert colors, raise contrast or apply a black and white threshold from 0 to 255. You can also limit or exclude characters. Each word gets an engine confidence score, which is a guide for review, not a measured accuracy rate.
Common mistakes to avoid
Do not scale a large image up without cropping first. A big source at 3x can pass the 16 megapixel limit, and the tool will ask you to reduce the scale or crop a smaller area. Source images can be up to 32 megapixels.
Remember that settings apply to every image in the batch, and changing them clears earlier results. Light text on a dark background usually reads better after using Invert colors. A threshold set too high can erase thin strokes, so preview the prepared image before extracting.
Useful tips and limitations
- Clear, upright printed text usually works better than handwriting or blurry screenshots.
- Confidence is an engine estimate, not a measured accuracy percentage.
- Choose Single line when you crop to one line of text, such as a serial number, because it stops the engine from looking for extra blocks.
- Turn on Preserve spaces between words when the spacing in a table or form matters to you.
Common questions
Are my pictures uploaded when I use Image to Text?
No. OCR runs on your device. The browser downloads self-hosted recognition assets and language models, but your images are not uploaded to our server.
What image formats and languages does the OCR tool support?
PNG, JPEG, WebP, BMP, AVIF and the first frame of GIF, subject to browser decoding support. Recognition languages are English, Spanish or both. SVG and PDF input are not accepted here.
Can I extract text from a PDF with this image OCR editor?
Use the separate PDF Tools OCR option for PDFs. This editor accepts image files, with a limit of ten images, 15 MB per image and 50 MB per batch.
Why does OCR need manual correction?
Image quality, fonts, handwriting, layout and language affect recognition. Crop and preprocessing can help, but always compare important names, dates and numbers against the original image.
Can I limit OCR to certain characters?
Yes. The character whitelist and blacklist fields accept up to 300 characters each, and a Digits only preset fills the allowed list with 0 to 9.
What do the page layout options do?
They tell the engine how text is arranged: a full page, one block, scattered text, a single line or a single word. Matching the layout to your image usually improves accuracy.
What is inside the per-image bundle?
It is a ZIP file with a TXT file of recognized text and a TSV file of word data for each image, which is useful for checking positions and confidence.
Your input is processed in your browser without a ToolWorld Online account. Copy or download anything you want to keep before leaving the page.