How to translate a PDF from English to Russian

In Translate PDF, change Target language from English to Russian. Only the text layer is translated: run a scan through OCR PDF first, and the translation is printed on a paper-coloured patch over each original line.

Open Translate PDF, upload the English document, and check two fields before you start: Source language and Target language. The tool replaces the text on each page with its translation and tries to keep every block where it was, but it only works with text that can be selected in the PDF.

This guide covers the steps, what happens with scanned pages, what the warnings mean, and what to check in the finished file before you send it on.

Check whether the file contains text

Open the PDF in any viewer and try to select a sentence with the mouse. If individual words highlight, the file has a text layer and can go straight to translation. If the whole page highlights as one picture, or nothing highlights at all, you are looking at a scan or a photo.

In a mixed file, check each part separately: the main text may have been typed in an editor while an appendix was scanned. The translator handles this as follows:

• if scanned pages are as many as text pages or more, the job fails with a message that the PDF has no text layer;
• if there are fewer scanned pages, the text pages are translated and the image pages come back unchanged, still in English.

Text inside pictures is not translated on text pages either: diagrams, screenshots, logos, and lettering in photos stay as they are.

Pick the languages: the default target is English

This is the most common trap. The Target language field starts out set to English. If you leave it, an English document gets "translated" into English and you receive almost the same file.

1. Upload the PDF into the tool. 2. In Source language, choose "English". "Detect automatically" also works for ordinary prose, but an explicit choice is safer for short labels and documents that mix languages. 3. In Target language, choose "Russian". 4. Click Process file and wait for the result. 5. Download the PDF and save it under a name that differs from the original, for example with an "RU" suffix.

Besides English and Russian, the lists include German, Spanish, French, Hindi, and Portuguese.

What happens to the layout

The translator removes the original text of a block and writes the translation into the same area of the page. Colour, approximate font size, bold and italic styling, alignment, and the font style (serif or sans serif) are carried over from the original. Images, lines, and table borders are left alone.

Russian text usually runs longer than English. If a translation does not fit the original box, the font of that block shrinks until it fits. In narrow columns, cells, and footnotes the letters can therefore end up noticeably smaller than their neighbours. Check the legibility of those spots separately, zoomed in.

Scans: run OCR first, then translate

The translator cannot read a scan that has no text layer, so first recognise the file in OCR PDF with English as the OCR language. OCR leaves the page image as it is and adds an invisible text layer to it (see the guide on recognising text in scans). Then upload the recognised PDF to Translate PDF and pick the languages as described above.

The English letters on a scan are pixels of the picture, so they cannot be removed like text. Instead, the translator covers each recognised line with a patch in the colour of the paper (sampled from the pixels around the line) and prints the translation on it in black, or in white on a dark background. The Russian text does not land on top of the English.

Keep in mind:

• The patch is a solid rectangle the size of the line or paragraph. It also hides whatever lay under the text in the picture: underlines, stamps, signatures, table rules.
• On an uneven background, such as stained paper, a coloured box, or a photo, the rectangle may stand out as a slightly different shade.
• The original text colour is not kept: on a scan the translation is always black or white.
• The result is only as good as the OCR. Recognition errors carry over into the translation, and anything OCR did not pick up (handwritten notes, stamps, lettering in diagrams and photos) stays in English in the picture.
• Scanned pages that have not been through OCR are not translated; the rules from the first section apply.

If you only need the text and not the layout, there is another route: extract the recognised text to TXT with PDF to Text and translate it in any text translator. The formatting is lost, but nothing is hidden under patches and the text is easy to proofread next to the page image.

Warnings and errors

Sometimes the translation service returns an empty answer for an individual block. That block stays in English, and the result carries a warning with the number of untranslated blocks. If half of the blocks or more could not be translated, the job produces no file and ends with an error instead; try running it again later.

A single run has a limit on the number of text blocks, currently 900. A block is a piece of text with one style: a paragraph, a heading, a caption, a table cell. A long report with many tables can hit that ceiling. Split it into parts with Split PDF and translate each one separately.

If the result shows empty squares instead of Cyrillic letters and there is a warning about a missing font, the problem is on the service side, not in your file: report it to support and try again later.

The translator cannot open a password-protected file. If you know the password, remove the protection first with Unlock PDF.

Check the translation against a concrete example

Suppose an instruction reads "Store at 5 °C". The Russian result must keep the number 5 and the unit °C, and the meaning must still be about storage. A mistake in a short instruction can matter more than clumsy style in a long paragraph.

Next, compare names, part numbers, and model identifiers. A label such as AB-120 should be carried over as an identifier rather than translated piece by piece. Pay close attention to decimals, dates, and negations: a dropped "not" turns a prohibition into permission. Russian uses a decimal comma, so check that a table does not mix "2.5" and "2,5".

Translation does not change the page count, so page-by-page comparison is easy: open the original and the translation side by side and scroll them together.

Look through every page at normal size, then zoom into tables with important values. Watch for three typical signs of trouble: English phrases among the Russian text, a squeezed tiny font, and lettering overlapping pictures. Read table headers and chart axis labels separately as well: machine translation handles short phrases without context worst of all, and an error in a column header changes the meaning of every value below it.

When machine translation is not enough

Machine translation helps you understand a letter, a manual, or an article. For a contract, a medical report, a technical regulation, or any other document that decisions depend on, the translation needs to be checked by a person who knows both languages and the subject. The tool does not guarantee legal accuracy of the wording and gives the translation no official status.

Keep the English original next to the Russian copy, and send the recipient both files if they may need to cross-check. Before sending, look separately at the last page and the appendices, which are the parts most often skipped during review.