How to Translate a PDF — What Works and What to Expect

PDF is the format translation tools like least: the text is frozen into a layout that translated text can't fit back into. Here are the methods that work, what each one actually delivers, why scans need OCR first — and when only a human translator will do.

← Back to Blog

How to Translate a PDF — What Works and What to Expect

Why PDF Translation Is Uniquely Awkward

Translating a Word file is easy: the text reflows. A PDF is different — it stores text as positioned runs frozen into a fixed layout, with line breaks, columns, and boxes built around the exact width of the original words. Translated text is almost never the same length: German runs roughly 30% longer than English, French 15–20% longer, Chinese dramatically shorter. The new text physically cannot occupy the old layout.

Fonts pile on a second problem: the PDF's embedded fonts may simply not contain the glyphs the target language needs — a document set in a Latin-only font has no way to display Japanese. So every PDF translation tool must choose between rebuilding a best-effort imitation of the layout (impressive when it works, messy when it doesn't) or producing a clean new document that prioritises the content over the artwork. Knowing which choice a tool makes tells you what you'll get.

Method 1 — Google Translate (Free, Best-Effort Layout)

translate.google.com has a Documents tab that accepts PDF uploads and returns a translated document approximating the original layout. It's free and fast, and for simple documents surprisingly serviceable. Expect rough edges on anything complex — multi-column text, tables, and text inside graphics come back scrambled or untranslated — and treat it as unsuitable for anything confidential (the document is processed by Google under consumer terms).

Method 2 — The Word Round-Trip (Editable Result)

When you need a result you can edit and reflow: convert the PDF to Word, translate the Word document (Word's built-in Review > Translate, or any translation tool), fix the layout in Word where the longer text broke it, and export back to PDF. It's the most work, and conversion imperfections stack with translation imperfections — but it's the only free path to a properly typeset translated document you control.

Method 3 — DeepL (Best Quality, Paid for Documents)

DeepL consistently rates among the best machine-translation engines for European languages, and its document translation preserves formatting well — but full PDF document translation sits in the paid tiers, with file and volume limits by plan. If translation quality of the prose is the priority and you translate regularly, it's worth the subscription.

Method 4 — Mapsoft PDF Hub (Clean Text-Focused PDF)

Our Translate PDF tool makes the other design choice, deliberately: it extracts the text, translates it chunk by chunk with an AI model, and builds a clean, plain-text PDF of the translation. The original layout, images, and fonts are not preserved — and that's the point: you get a readable, complete translation of the content without a mangled imitation of the design. If any section fails to translate, the output says so rather than silently omitting it. Free to try; included with Super User; the extracted text is processed via the OpenAI API (details in the Hub's privacy policy). It pairs well with the Summarize tool, which can translate just the summary when you only need the gist.

Scanned PDFs: OCR First

Translation engines need text, and a scan is a photograph. Run OCR first — Acrobat's Recognize Text or the PDF Hub OCR tool — then translate the OCR'd copy. Be aware the errors compound: a character OCR misreads becomes a word the translator misunderstands. For important scanned documents, skim the OCR text before translating it; our OCR guide covers getting recognition quality up.

When Machine Translation Isn't Enough

Machine translation answers "what does this document say?" superbly. It does not produce:

  • Certified translations. Immigration, courts, universities, and registries require translations produced and attested by a qualified human translator. No machine output qualifies, whatever the quality.
  • Legally reliable contracts. A mistranslated obligation is still an obligation. Use machine output to understand a contract; have a professional translate one you'll sign.
  • Publication-ready copy. Marketing and documentation that represents your brand deserves a human pass — machine translation reads as competent, rarely as native.

Tool Comparison

ToolResultPricingBest for
Google Translate (Documents)Best-effort layout imitationFreeQuick gist of simple documents
Word round-tripFully editable document you re-typesetFree with WordWhen you need a proper, editable result
DeepLFormatted document, top prose qualityPaid tiers for PDF documentsRegular high-quality European-language work
Mapsoft PDF Hub TranslateClean plain-text PDF of the translationFree to try; included with Super UserReadable full-content translation, no layout pretence
Human translatorCertified / publication-gradePer-word professional ratesAnything official, legal, or brand-facing

Related Articles

How to Convert PDF to Word

The first leg of the round-trip method — get an editable document out of the PDF.

How to Summarize a PDF with AI

Sometimes you don't need the whole document translated — just the gist, in your language.

OCR in Adobe Acrobat

Scanned documents need a text layer before any translation engine can touch them.

Translate a PDF Now

Use Mapsoft's PDF Hub to translate, summarize, and chat with PDFs — plus merge, split, compress, and 50+ more tools, directly in your browser.