Russian image to text turns Cyrillic in a picture into text you can edit, search and paste. Typical jobs: a chapter from a scanned book for a paper, an article from a PDF that won’t let you copy, a certificate or letter you need typed out for a translation, a screenshot from Telegram, a recipe card, or a page of instructions.
The page starts with Russian and English selected, so documents with English names, terms or references are read in one pass, along with product codes, web addresses and email addresses.
Cyrillic and Latin look-alikes
Cyrillic and Latin share many letter shapes. In lowercase, а е о р с у х look exactly like a e o p c y x, and the capitals А В Е К М Н О Р С Т Х have Latin twins too. On screen they’re identical. To a computer they’re different characters.
That matters more than it seems. A Russian word with one Latin letter inside looks perfect, but it won’t be found by search, it gets flagged by spell-check, and it sorts in the wrong place in a list. When both Russian and English are selected, short words, abbreviations and words in capitals are where a Latin letter is most likely to slip in, because they give the engine the least context.
Two ways to deal with it:
- Read with Russian only. If the page has no English, untick English in the language menu (the globe). The page is read again when you close the menu, and every letter comes from the Cyrillic model.
- Find the stray letters. Open find and replace with Ctrl+F (⌘F on a Mac), tick regexp and search for
[A-Za-z]. Every Latin letter in the text is highlighted, so in a Russian-only document anything that lights up needs fixing.
Letters that are easy to misread
Beyond the look-alikes, a few Cyrillic letters are close enough to each other, or to digits, to cause regular mistakes:
- ш and щ, and и and й, which differ by a small tail or mark
- е and ё, since the two dots are tiny and many texts print ё as е anyway
- ь, ъ and ы, which are easy to confuse at small sizes
- б and 6, з and 3, ч and 4, especially in tables and forms where numbers and letters sit side by side
Italic text deserves extra care. In Russian italics, т looks like a Latin m, и like u, п like n and д like g, so italic passages, captions and emphasized words can come out with very different letters from the image. Words the engine was unsure about are underlined, and a button jumps from one to the next. Click any line and its place lights up on the image.
Ukrainian, Bulgarian and other Cyrillic languages
Ukrainian and Bulgarian have their own options in the language menu. Choose the one that matches the page: Ukrainian uses і, ї, є and ґ, which Russian doesn’t, and both languages spell words differently enough that the right model makes a real difference.
Serbian, Belarusian, Kazakh, Macedonian and other Cyrillic languages aren’t in the list. Russian is the closest choice, but letters unique to those languages, such as ј, љ, ў or қ, are likely to come out wrong, so treat the result as a draft.
Old books and pre-reform spelling
Printed books, including Soviet-era editions, are generally a good fit: the type is regular and the lines are even. The quality of the photo or scan decides most of the result.
- Press the book flat and photograph one page at a time, since lines curve near the spine. See photo to text for more on photographing pages.
- Lift yellowed or speckled paper with Contrast, Black & white or Reduce noise. Each page can be adjusted and read again on its own.
- Crop to one column in journals and newspapers so neighboring columns don’t run together.
Books printed before the 1918 spelling reform use letters such as ѣ, і, ѳ and a hard sign at the end of many words. Expect those to need more correcting than modern text.
Tables and numbers
Russian writes numbers with a decimal comma and spaces between thousands, as in 1 234,56. Numbers in that form come through as they’re printed. If you take a table into a spreadsheet, check those columns: a value like 12,500 can be taken as twelve thousand five hundred rather than twelve and a half, and others may stay as text. See image to Excel for working with tables.
Getting a cleaner result
Clear printed Cyrillic is usually read very accurately. Most problems come from the image: low resolution, shadows, blur, an angle, or text that’s too small. Move closer rather than zooming, straighten angled photos with four corners, and crop to the part you need. For a fuller list, read how to get accurate OCR results.
Where the text goes next
The result is ordinary Unicode Cyrillic, so it pastes into Word, Google Docs, email or Telegram like typed text. Rich copy keeps headings and lists. You can also download:
- Word (.docx) for editing. See image to Word.
- Searchable PDF, which keeps the scan exactly as it was and adds an invisible text layer, so you can search an old book with Ctrl+F
- PDF, which for Cyrillic opens your browser’s print dialog, where you choose “Save as PDF” so every letter is drawn correctly
- Plain text, Markdown or HTML
For a long PDF or a stack of page photos, every page is read in turn, and Whole document view combines them into one file. See PDF to text.
Private by default
Personal letters, certificates and documents are read in your browser, on your device, and aren’t uploaded. History is off unless you switch it on, and then it stays in your browser. Read more in is online OCR private?