Extracting Text with Offline OCR
One tap turns a photo, a screenshot or a scanned page into text you can copy, search or save - read on the phone itself, with no picture ever leaving the device.
Why You'll Love This
A shop sign in another language, a recipe photographed from a magazine, a scanned contract page, a screenshot of an error message you want to search for - printed words trapped in a picture are everywhere. FastMediaSorter reads them for you with an on-device OCR engine, so the words become text you can select, copy, search or save. Nothing is uploaded anywhere: the recognition runs on the phone itself.
The engine is Tesseract, tuned here for strong Cyrillic support - so a Russian or Ukrainian sign is read as Russian or Ukrainian letters, not as the nearest-looking Latin ones.
- FastMediaSorter in an edition with OCR: Standard, noLegal, Legacy or VR. The Lite, Photos and FOSS editions do not read text from pictures.
- The OCR Engines extension, downloaded once from Settings, the General tab, Downloadable Extensions - or turned on right on the setup wizard's functionality page. See first launch and setup and downloadable extensions.
- For better recognition of Cyrillic text: the Russian OCR Model or the Ukrainian OCR Model, downloaded the same way.
- A device with Android 8.0 (API 26) or newer and at least 3 GB of RAM. On an older or smaller device you will see "OCR is unavailable on this device."
- A picture with legible text: a photo already open in the app, a fresh photo from the camera, a screenshot, or a scanned PDF page.
Pull text out of a picture you are viewing
Open the picture in the image viewer or the reader, show the command panel, and tap the OCR button (its accessibility label is Extract Text). The recognized words open on a Recognized Text screen as plain, selectable text - copy all of it, copy a piece, or share it like any other text.
If the picture has no legible words - a blurry photo, a plain landscape - you will see "No text found in image".
A photographed sign open in the image viewer with the OCR result panel showing the recognized words as selectable text.
Snap a photo and get instant text
For a fresh subject - a menu, a poster, a business card - skip the gallery: open the Camera OCR program from the programs menu, drop a Camera OCR widget on your home screen, or use the Capture & OCR panel item, then point the camera and shoot.
After "Processing photo.." a crop frame appears over the photo. Drag its corners around the exact block of text, tap Retry to shoot again, or tap OK to continue. While "Extracting text and translating.." runs, the result screen opens with an Original text tab and a Translation tab, plus Save as .txt to write the text to a file and Next photo to shoot the next one without leaving the flow.
A just-taken photo of a shop sign with the Camera OCR crop frame open, corner handles visible, Retry and OK buttons at the bottom.
Turn a screenshot into text
Already looking at something with text on screen - a chat, a web page, a form - and would rather not leave the app? In the Standard and noLegal (sideload) editions, open the capture strip and tap Screenshot - OCR translation: the screenshot opens straight in the same crop-and-recognize screen as a photo, and only the cropped piece is kept in your gallery. The same strip's Take a photo and OCR-translate action does the same thing starting from the camera instead of the screen.
On Legacy and VR, where the capture strip is not available, the Camera OCR program and its widgets cover the same ground - recognition itself works the same wherever OCR is available.
Choose the right recognition language
Recognition quality depends on picking the right alphabet. Tap the OCR language or Translation language row - in the Camera OCR crop screen, or in the reader's Text Settings dialog - to open a searchable list of every language the engine currently supports, ordered by your interface language, with Auto available for the source language.
For the best result, pick the source language explicitly rather than leaving it on Auto, especially for Cyrillic text: guessing reads the picture with the English alphabet first, and a look-alike Latin letter can slip in where a Cyrillic one belongs. Choosing Russian or Ukrainian up front reads those letters correctly from the start.
The searchable language picker open over the Camera OCR crop screen, search field focused, Russian highlighted as the current selection.
Recognize text on a scanned PDF page
A PDF made from a scanner or a photocopy has no text layer at all - the words are just part of the picture. Open the page in the reader and tap Translate; a picture-only page goes through the same offline OCR pass before its recognized words are shown. See reading PDF documents for finding the Translate button, and translating extracted text for what happens to the recognized words next.
Copy or save what was found
From the Recognized Text screen or the Camera OCR result, select all or part of the text and copy it like any other selection, or tap Save as .txt to write it to a timestamped text file in your chosen destination folder - you will see "Saved text to.." with the path when it lands.
What You Get
Text trapped in a photo, a screenshot or a scanned page becomes real, selectable words - copied, searched, or saved as a text file, entirely on the phone. Camera OCR turns a fresh photo into text in a couple of taps, the crop screen works the same way for a screenshot, and choosing the right language keeps Cyrillic text from turning into Latin look-alikes.
Tips and Troubleshooting
- Text recognition works in the Google Play version too. Its engine now ships inside the app, so a Play install no longer shows a misleading message that it is not available.
- The crop frame is easy to grab. The bottom handle of the crop frame stays clear of the buttons at the bottom of the screen, and the frame's border is drawn in full.
- Want the translated text side by side, not just the original? See translating extracted text.
- Blurry photo or a loose crop? OCR only reads what is actually printed - a steadier photo and a tighter crop around the text read better than a wide, blurry one.
- Only want the language packs, not a tour of the whole extensions screen? Jump straight to downloadable extensions and language models.
- Setting up the app for the first time? Turn OCR on right on the functionality page - see first launch and setup.