How to Translate a German PDF Book Into English

Translating a book-length PDF has two separate problems, and conflating them is why the results are usually disappointing. The first is getting the text out. The second is translating it well. A tool that does both in one step tends to do neither properly.

First, find out what kind of PDF you have

Try to select a paragraph. If text highlights, the file has a real text layer and you can work with it directly. If nothing selects, it is a scan — a picture of pages — and no translator can read it until OCR adds a text layer.

This matters most for older German books, where two further complications are common: Fraktur (blackletter) type, which general OCR reads very badly, and the pre-1996 orthography, which spellcheckers and some translation models handle inconsistently.

For scanned books: OCR with the right language setting

Set the OCR language to German before running it — recognition accuracy drops sharply with the wrong language model, because the engine uses language statistics to disambiguate similar glyphs. ABBYY FineReader has specific Fraktur support and is the most reliable for older texts. OCRmyPDF (built on Tesseract) is free and local, and Tesseract has a deu_frak model for Fraktur, though it is noticeably less accurate than ABBYY on poor scans.

Note that Online PDF Edits does not perform OCR — this step has to happen elsewhere.

Then translate

DeepL is generally regarded as stronger than the alternatives on German prose specifically, particularly with compound nouns and subordinate clause order, and it accepts document uploads that preserve basic formatting. Google Translate handles documents too and is free at higher volumes. Both have upload size limits that a full book will exceed, so plan to split the file.

For book-length work, split into chapters first — extract pages does this — and translate chapter by chapter. It keeps you under upload limits and makes it far easier to check quality as you go.

Expect to review, especially for anything that matters

Machine translation of literary or academic German is serviceable for comprehension and unreliable for publication. Idiom, register and long subordinate constructions are where it slips. If the translation is going to be quoted or published, budget for a human pass — and check for the compounding problem where an OCR error becomes a confident mistranslation, which is very hard to spot in the output alone.

Frequently asked questions

Can I translate a PDF without losing the layout? Partly. DeepL and Google both attempt to preserve document structure, but German runs roughly 10–30% longer than English in some constructions and shorter in others, so text no longer fits its original boxes. Expect to fix layout afterwards.

What if the book is in Fraktur? Use OCR with an explicit Fraktur model — ABBYY, or Tesseract's deu_frak. General-purpose OCR reads blackletter very poorly and will produce nonsense that looks like text.

Is it legal to translate a book? Translation is a restricted right under copyright. For personal study it is usually fine; distributing a translation of an in-copyright work needs permission. Public-domain texts are unrestricted.

How do I handle a 500-page book? Split it into chapters, OCR and translate each, then merge. Every tool involved has limits a whole book exceeds.