You receive an important document in a language you don’t understand. A technical manual in German, a contract in Italian, a research paper in Japanese. You need to know what it says, but hiring a professional translator for the whole document is too expensive, or you simply need a quick translation to grasp the general content.
Machine translation tools have improved enormously in recent years. Google Translate, DeepL and others can provide decent translations of entire documents for free. They aren’t perfect; they may contain errors or sound a bit odd, but they’re usually sufficient for grasping the general gist. The trick lies in using the right method to retain some of the original formatting and achieve the best possible quality.
Google Translate for PDFs
Google Translate can translate PDF documents directly. Go to translate.google.com, select the ‘Documents’ tab instead of ‘Text’, upload your PDF, choose the source and target languages, and Google processes the document.
The result is a translated document that you can download. The translation is usually reasonably good, especially for common languages such as English, Spanish, French and German. For less common languages or those with very different structures, the quality drops.
The main problem is the formatting. Google Translate does not retain the original layout of the PDF. The translated document is plain text with basic formatting. If your original PDF has a complex layout, tables, or images containing text, all of that is lost or becomes disorganised.
For documents where only the content matters and not the formatting, Google Translate works perfectly and is completely free with no limits. For documents where the layout matters, you need other methods.
DeepL for better quality
DeepL has a reputation for producing higher-quality and more natural translations than Google Translate, especially for European languages. The free version allows you to translate documents up to 5 MB.
The process is similar: you upload the PDF, select the languages, and DeepL translates. It retains slightly more of the original formatting than Google, but it isn’t perfect either. Complex tables and elaborate layouts are simplified.
DeepL only supports certain languages. If you need to translate to or from languages that DeepL does not cover, Google Translate offers much broader support.
Convert to Word, translate, convert back
To preserve the formatting better, you can convert the PDF to Word first, translate the Word document, and then convert it back to PDF. This method requires more steps but the result retains more of the original layout.
Convert your PDF to Word using any converter. Open the resulting Word document and use Microsoft Word’s built-in translation feature (if you have Office 365) or copy the text into Google Translate or DeepL. Adjust the formatting if necessary and convert it back to PDF.
It’s more work, but you have control over the process. You can review the translation, correct obvious errors, and adjust the formatting. For important documents where you want a presentable result, this method is worth it.
Copy and paste text
For short PDFs or when you only need to translate specific sections, simply copying the text from the PDF and pasting it into Google Translate or DeepL is the quickest option.
Open your PDF, select the text you need to translate, copy it, paste it into the translator, and get the translation. You don’t have to upload entire files or worry about limits.
This method doesn’t work well with scanned PDFs where the text is an image, not actual text. Nor does it retain any formatting. It’s purely for extracting textual content and translating it.
Scanned PDFs require OCR first
If your PDF is a scan (images of pages, not actual text), translation tools cannot read the content directly. You first need OCR to convert the images into text.
Google Drive can do this. Upload the PDF to Drive, open it in Google Docs, and Google automatically applies OCR, converting the document into editable text. You can then use Google Translate to translate it.
The quality depends on the clarity of the original scan. Documents that are clearly scanned are recognised well. Blurry, smudged or poorly lit scans produce poor results full of errors.
Online PDF translation services
Several specialised web services translate PDFs whilst preserving the formatting better than general-purpose translators. OnlineDocTranslator, DocTranslator and similar services use Google or Microsoft translation engines but attempt to preserve the original layout.
You upload your PDF, choose the languages, the service processes it and returns a translated PDF with a layout similar to the original. Results vary depending on the document’s complexity but are generally better than using Google Translate directly for documents with a layout.
These services usually have limits in their free versions: maximum file size, number of pages, and translations per day. For large documents or frequent use, considerpaid versions.
Limitations of machine translation
No machine translator is perfect. They all make mistakes, especially with: specific technical language, idioms and phrases, wordplay, proper nouns that should not be translated, cultural context, ambiguities in the original language.
Read the result with a critical eye. If something sounds strange or doesn’t make sense, the translation is probably incorrect. For important documents where accuracy is critical, use machine translation as a first draft and have it checked by someone who knows both languages.
Do not use machine translations directly for legal documents, contracts, medical documents or anything where errors could have serious consequences. In these cases, invest in a professional human translator.
Batch translation
If you need to translate multiple PDFs, doing them one by one is tedious. Some services allow you to upload several documents simultaneously or process them sequentially with the same settings.
This is useful if you have, for example, several chapters of a manual or multiple research articles in the same language that you all need to translate into your own. You set it up once and process the whole batch.
Keeping PDFs in both languages
It is often useful to keep both the original and the translation. The original as a reference if you are unsure about something in the translation, and the translation to read and understand the content.
Consider combining both PDFs into a single document where alternate pages are the original and the translation. Or keep them separate but clearly labelled. Organise your files so that you can easily find which document is which.
Translate only specific sections
You don’t always need to translate the entire document. If you’re only interested in certain sections, translate just those. It’s quicker and avoids wasting time on irrelevant parts.
You can extract specific pages from the original PDF, translate only those pages, and keep the rest in the original language. Or simply copy and translate the relevant paragraphs without touching the rest of the document.
Quality depending on language pair
Translation quality varies greatly depending on the language pair. English-Spanish, English-French, and English-German usually translate quite well because they are common languages with plenty of training data.
Less common languages or those with very different structures (Japanese-Spanish, Arabic-English, Finnish-Portuguese) produce more problematic translations. The tools have less data for these pairs and structural differences complicate accurate translation.
If the initial translation is very poor, try translating first into English (which usually works better) and then from English into your language. Two translation steps accumulate more errors but sometimes the final result is more comprehensible than a problematic direct translation.
Check with native speakers
If the translation is for an important purpose, ask someone who speaks both languages to review it. Machine translation can produce text that is technically correct but sounds strange or uses expressions that are inappropriate for the context.
A native speaker will quickly spot errors that a non-native might not notice. This is particularly important for documents you intend to share publicly or use professionally.
Privacy when translating
When you upload a PDF to online translation services, you are sharing the content with those services. For public documents or general information, this is not a problem.
For confidential documents, private contracts or sensitive information, consider the privacy implications.
Read the services’ privacy policies. Google and DeepL state that they do not use the content of translated documents to train their models, but your files do pass through their servers. If this is unacceptable, you need local solutions, even if they are more limited.
Alternatives for professional translations
If you need a flawless translation with a quality guarantee, nothing replaces a professional human translator. Platforms such as Gengo, Translated or local translators offer professional translations. They cost more but the quality is incomparably superior.
For legal, medical or specialised technical documents, or anything where errors could cause problems, invest in professional translation. Machine translation is a useful tool for understanding general content, not for creating critical final documents.
Automatically translating PDFs is an incredibly useful resource for understanding documents in foreign languages quickly and at no cost. Current tools are surprisingly good for general translation. But be aware of their limitations, review the results critically, and turn to professionals when accuracy is essential. Machine translation gives you quick access to content; human translation gives you accuracy and nuance that machines cannot yet match.