What is PDF translation?
PDF translation is the process of converting the text in a PDF file from one language into another and delivering a PDF that still looks like the original: same pages, same columns, same tables, same images, same fonts wherever the target script allows it. The translation of the words is only half the job. The other half is getting those words back into a fixed layout that was never designed to change.
That second half is why PDFs have a reputation as the hardest everyday file to translate. A Word document is a flow of editable text. A PDF is closer to a printed page: every line of text is placed at an exact position, and the file often has no idea which lines belong to the same paragraph, which cells belong to the same table row, or whether a block of text is a heading, a caption or a footnote. Smartcat’s PDF Translator describes the problem plainly: most translators pull the words out and hand back plain paragraphs, so tables collapse, headers and footers disappear, and two-column pages merge into one. Someone then rebuilds the file by hand — once per language.
Where this guide fits
PDF translation is one part of document translation. If you work with Word, Excel, PowerPoint or other office files as well, start with our Document translation guide, which covers every format and workflow. This guide goes deep on PDFs only.
What a good PDF translation gives you
A usable translated PDF has four properties. The text is accurate and uses consistent terminology. The layout survives: tables, columns, headers and footers stay where they were. Images and diagrams stay in place, ideally with the text inside them translated too. And the file is ready to send, so no designer has to reset it in a desktop publishing tool. Smartcat pulls the text out of your PDF, translates it into any of 280+ languages, and rebuilds the document with the original layout, fonts, tables and images in place, with built-in OCR for scanned files.
Who searches for a PDF translator
Two very different people type “translate PDF” into a search box. One has a single file — a letter, a manual, a research paper — and wants to read it in their own language today. The other runs a team that ships PDFs across markets every month: product sheets, SOPs, catalogs, compliance documents. Free tools are often fine for the first person. The second needs layout fidelity, a glossary that holds across files, human review where the stakes are high, and security. This guide covers both, and is honest about where each kind of tool fits.
Digital, scanned and design-heavy PDFs
Before you pick a tool, work out what kind of PDF you have. The type of file decides which translation route will give you a clean result, how long it takes, and how much checking it needs afterwards.
Digital PDFs
A digital (or “native”) PDF was exported from a word processor, design tool or other application. You can select and copy its text. Because the text layer already exists, a translator can read it directly, translate it and set it back in place. Brochures, reports, whitepapers, user manuals and most business documents fall into this group. They are the best candidates for fully automated, layout-preserving translation.
Scanned PDFs
A scanned PDF is a picture of a page. Archived files, signed forms, faxed dossiers, old manuals and stamped certificates usually arrive this way. There is no text layer to read, so the text has to be recognised first with OCR (optical character recognition) before anything can be translated. Scan quality now matters as much as translation quality. We cover this in detail in the OCR section below.
Design-heavy PDFs
Catalogs, infographics, sales sheets and point-of-sale materials are light on text and heavy on visuals. Much of their text sits inside graphics, callouts and images rather than in neat paragraphs. These files need a route that translates text where it sits and rebuilds the design around it — which is exactly what Smartcat’s Translation Agent for design-heavy PDFs was launched to do.
Three PDF workflows, one decision
Smartcat’s help center documents three distinct PDF translation workflows and when to use each. The short version is in the table below; the full decision tree is in Choosing the Right PDF Translation Workflow in Smartcat.
| Workflow | AI PDF Translation | PDF via Image Pipeline | Standard PDF (OCR → Word) |
|---|---|---|---|
| Best for | Text-based PDFs where the design matters | Scanned PDFs, forms, handwritten content | Simple, text-heavy PDFs |
| Layout preservation | Same design as the original | Better than OCR → Word for scans | Approximate |
| Scanned PDFs | Not supported | Yes (choose pdfimageonly at upload) | Via OCR |
| Editing | Real-time preview, page redo | Full manual editing in the editor | Full CAT editor, editable Word file |
| Glossary & translation memory | Yes | Yes | Yes |
A practical rule from the same article: if you have a batch, sort it into a digital-native pile and a scanned pile and translate them separately, attaching the same glossary and translation memory to both.
Translating scanned PDFs with OCR
OCR turns the pixels of a scanned page into characters a translation engine can read. It is the step that decides whether a scanned PDF translation works at all: a misread character in the source becomes a confidently wrong word in the translation, and nothing in the output warns you. A misread digit in a torque specification reads exactly as convincingly as a correct one.
Smartcat’s scanned PDF translator is built for archived files, signed forms, old manuals and flattened pages. In practice, scanned files run through the image-translation pipeline: you choose pdfimageonly as the parsing method when you upload, and each page is treated as an image, recognised, translated and rebuilt. According to Smartcat’s documentation, this route typically produces much better layout preservation for scans than a traditional OCR-to-Word conversion, and it is the route that supports handwritten content.
How to get a clean scanned-PDF translation
Start with the best scan you can get
Aim for at least 300 DPI, with pages straight, evenly lit and high in contrast. Clean scans of typed text translate reliably. Low-resolution scans, handwriting and dense multi-column tables are where OCR errors cluster.
Pick the image route at upload
For scanned pages, select pdfimageonly in the upload dialog so each page goes through the image pipeline. The AI PDF workflow for digital files does not handle image-of-text pages.
Spot-check the middle, not just page one
In a long scan, errors hide in the pages nobody checks. Review a sample from throughout the file, and give numbers, names, dates and tables a second look.
Add a human pass where errors carry risk
For contracts, certificates, medical records or safety documentation, route the high-stakes sections to a professional reviewer rather than trusting a single machine pass.
OCR quality has improved a lot. In 2025 Smartcat upgraded its PDF processing engine for more precise text extraction, retained tables and vector graphics, and better layout preservation. Files uploaded before the change are not reprocessed automatically, so re-upload older PDFs if you want the newer engine. For a practical walkthrough of the common problems — unreadable text, handwriting, mixed layouts — read Translate scanned PDF: easy steps and best practices.
Scanned PDF or image?
A scanned page is technically an image, so scanned PDFs overlap with photo and image translation. If your source is a set of photos, screenshots or JPG scans rather than a PDF, the Image translation guide covers that workflow, and Smartcat’s scanned document translator handles single scans directly.
How to translate a PDF and keep the layout
Keeping the layout is the main reason people look for a dedicated PDF translator rather than copying text into a general translation box. Three things work against you.
1. Text expansion and contraction
Translations rarely take up the same space as the source. English typically runs 20–60% longer than Japanese and 30–60% longer than Chinese, according to Smartcat’s Japanese and Chinese PDF translator pages. Going the other way, English is around 15–20% shorter than French. In a fixed PDF layout, longer text overprints, truncates or disappears, and shorter text leaves gaps. A good PDF translator reflows text boxes, tables and columns to absorb the difference instead of patching each line.
2. Tables, columns and structure
Because a PDF stores text as positioned fragments, a naive extractor reads across columns and merges table cells into run-on lines. Smartcat rebuilds the document structure rather than re-typing text, so columns, tables, headers and footers come back in place. Its help center is candid that complex tables and multi-column layouts may still need minor alignment adjustments after translation, so check them before you send the file on.
3. Images, diagrams and text inside pictures
Many PDFs carry meaning in their visuals: a diagram’s callouts, a chart’s axis titles, a screenshot’s button labels. When you translate a PDF with pictures in Smartcat, images, logos, icons and vector diagrams are detected and recreated in their original positions, and text baked into an image is recognised, translated and written back into the picture. Figure captions stay attached to their figures. For brochures, signage, product sheets and visual documentation, the PDF image and text translator handles image text and page visuals in one flow, with a visual editor for review in context.
Fonts and right-to-left scripts
Where the original font supports the target script, the output stays close to the source. Where it does not — a Latin display font asked to render Japanese, for example — a similar font is substituted. Bold and italic are generally preserved, but exact font matching isn’t guaranteed. Some complex right-to-left layouts (Arabic, Hebrew) may need manual adjustment, and interactive elements such as form fields and buttons may not be preserved, so test any fillable PDF before you distribute it.
Layout checklist before you export
- Translate a single test page first for complex documents, to catch layout issues early.
- Check pages in long target languages (German, Finnish) for overflow in tables and text boxes.
- Review in-image text, which can be slightly less accurate than body text.
- Confirm that headers, footers, page numbers and captions are where they should be.
- If you need to keep editing the content, export the editable working file as well as the PDF.
Translating large PDFs and batches of files
Large PDFs fail in most translation tools in one of two ways: the upload stops at an unstated size cap, or a free tool hits its page limit and asks you to split the file. Splitting is the expensive failure. Translate a 380-page manual in separate runs with no shared memory and the same part name comes back three different ways in three different chapters.
Smartcat’s large PDF translator takes a different approach. You upload the whole file — the AI PDF workflow has no page limit and accepts files up to 100 MB — and Smartcat segments it internally, translating every segment against one shared translation memory and glossary. A term rendered on page 4 is reused verbatim on page 400. Work can be paused, resumed and exported without resetting the project, and pricing is by words rather than pages.
What to expect on long documents
- Time: a typical 10–20 page document takes 15–45 minutes; documents of 60+ pages take significantly longer.
- Size: over 100 MB, the documented remedy is to split the file with a PDF editor such as PDF-XChange Editor or Adobe Acrobat and translate the parts.
- Review: in the editor you can filter and search segments and jump straight to warnings, disclaimers and spec tables instead of scrolling through 400 pages.
- Scans: scanned manuals use the image pipeline you select at upload. Low-resolution scans and handwriting stay unreliable at any length.
Batches of PDFs
If you translate a whole product line at once, upload the files into one project. Each PDF becomes a document in the same project, sharing a project-level glossary and translation memory, and progress is tracked per file so one problem document doesn’t hold up the rest. When the files are a mix of formats, not only PDFs, Smartcat’s large file translation covers Word, PowerPoint and other documents in the same way.
PDF translators compared: free tools, AI chat and Smartcat
There is no single best PDF translator; there is a best fit for the file and the stakes. Our comparison of the 10 best PDF translation tools tested them against what actually decides the outcome: layout preservation, language coverage, linguistic quality, OCR for scanned files, review and editing options, and data security. Its conclusion is a useful starting point: free tools are fine for simple, day-to-day PDFs, and OCR quality is the deal-breaker, so test every tool on your worst scanned file, not a clean one.
| What you need | Free online converters | General-purpose AI chat | Smartcat |
|---|---|---|---|
| What you get back | A rough file or copyable text | Translated text you rebuild yourself | A finished PDF in the new language |
| Layout, tables & images | Partial; formatting often lost | Text only | Rebuilt in place |
| Scanned PDFs | Limited | Reads many scans; output is still text | OCR / image pipeline, text back into the layout |
| Long documents | Per-file size and page caps | Context limits; long files get chunked | No page limit |
| Terminology across files | No | Glossary pasted into each prompt | Glossary & translation memory |
| Human review | No | Arrange it yourself | Vetted linguists in the same workflow |
The table summarises the comparison on Smartcat’s PDF Translator page, which compares named tools in detail.
Using ChatGPT or other AI chat to translate a PDF
Pasting a PDF into an AI chat works for getting the gist of a short document. It breaks down when you need a deliverable: the output is text in a chat window, the layout has to be rebuilt by hand, terminology isn’t stored between sessions, and output varies run to run. PDF translation with AI, beyond ChatGPT explains how to keep AI speed but move the work into a layout-safe workflow with reusable terminology and a real review path. If you’re weighing AI translation tools more broadly, the AI translator guide covers how they work and how accurate they are.
PDF tools for specific jobs
Translating PDFs in the browser
If you want to skip downloading and re-uploading files, Smartcat’s PDF translate extension for Firefox lets you translate documents directly in the browser, with the same layout-safe translation, visual review and export.
Turning PDF tables into translated spreadsheets
Sometimes the goal isn’t a translated PDF at all but the data inside it. Price lists, specification tables and reports are easier to reuse as spreadsheets. PDF to Excel translation extracts table data from a PDF, translates it and returns a structured Excel file with the table layout preserved.
When to use a PDF editor instead
Desktop PDF editors are the right tool for small fixes after translation: correcting a word, adjusting a line break, or splitting a file that is over the upload limit. They are not built for translating a document end to end into several languages with consistent terminology. Use the editor for the last mile, and the translation platform for the translation.
Translate PDFs by language
Smartcat supports 280+ languages for PDF translation, with one source language and as many target languages as you need in a single project. The highest-demand pairs have dedicated pages that explain the specific problems each language creates in a PDF.
Spanish
Spanish (Latin America) and Spanish (Spain) are separate source languages in Smartcat, along with country variants such as Mexico, Argentina, Chile, Colombia and Peru. English output is roughly 15% shorter than Spanish, so the rebuild redistributes the freed space. In the other direction, Spanish typically expands compared with English, which our step-by-step guide to translating a PDF to Spanish without losing formatting covers in detail.
German
German breaks PDF tools in specifically German ways. Long compound nouns overflow table cells, and weak OCR reads ß as B or flattens umlauts, which changes meaning. The German to English PDF translator applies your glossary and rebuilds tables and columns so they come out intact.
French
French PDFs tend to come back subtly wrong rather than obviously broken. The French to English PDF translator converts French punctuation to English conventions and rebalances columns and tables for the shorter English text.
Chinese
Chinese (simplified) and Chinese (traditional) are separate selectable source languages, each with regional variants. Because English runs much longer than Chinese, the rebuilt PDF reflows tables, columns and text boxes to absorb the expansion. Use the Chinese to English PDF translator for incoming documents and the PDF translator for Chinese and global languages when you need to translate PDFs into Chinese.
Japanese
A single Japanese sentence mixes kanji, hiragana and katakana with no spaces between words, so tools that split text on whitespace shred it. Smartcat segments Japanese linguistically and handles keigo consistency with custom prompts, AI translation profiles and your glossary. One limit is worth knowing up front: vertical (tategaki) typesetting is not supported, so re-export the original horizontally or route it to a human translator.
PDF translation use cases
Enterprises run on PDFs, and the same handful of document types come up again and again. Each one has a different balance of layout complexity, risk and volume.
Manuals, SOPs and technical documentation
Long, text-heavy, terminology-dense and often hundreds of pages long. Consistency is the whole game: one shared translation memory and glossary keep part names and warnings identical from the first page to the last. Upload the whole file through the large-PDF workflow described above and review the safety warnings and spec tables rather than every page.
Marketing collateral, catalogs and brochures
Design-heavy files where the visuals carry the message. Use the route built for image-heavy PDFs so the text inside headlines, callouts and graphics is translated in place, and check the result in the visual editor before export.
Contracts, legal and medical documents
High-stakes PDFs need more than a machine pass. Smartcat’s help center positions AI PDF translation of contracts and legal documents as being “for reference purposes”, and immigration offices, courts and universities usually require certified human translation signed by the linguist. Use AI for the fast working version and route the official version to a professional reviewer. Our Translation by industry guide covers the requirements for legal, life sciences and other regulated sectors.
Reports, whitepapers and research papers
Digital PDFs with charts and figures. The AI PDF workflow handles these well, and translating the labels inside charts keeps figures readable. For a single paper you just want to read, a quick AI translation is usually enough; for a report you will publish, apply your glossary and review the charts.
Archived records, forms and certificates
Usually scanned, sometimes stamped or handwritten. Use the scanned-PDF route, check numbers and names carefully, and add human review where the document has legal weight.
How to translate a PDF in Smartcat, step by step
The workflow is the same whether you translate one brochure or a folder of manuals. It is documented in full in Translating PDF Documents with AI.
Prepare your glossary and translation memory
Add key terms and their translations to your workspace glossary before you start. The AI applies them consistently throughout the document, and translation memory reuses anything you have translated before.
Upload the PDF
Drag the file into the Smartcat AI chat or create a project. For digital PDFs, the AI PDF Translation workflow is the default. For scanned PDFs, create the project first, then upload the file and choose pdfimageonly as the parsing method.
Choose your languages and add context
Pick one source language and as many targets as you need. You can tell the AI what kind of document it is or which industry it belongs to, to improve accuracy.
Watch the translation and review what matters
Each page appears in a live preview as it is translated. Ask the AI to redo a page or translate a page range, check the result against the layout in the visual editor, and add professional review from Smartcat’s marketplace of vetted linguists where the stakes demand it.
Download the finished PDF
Export a PDF that mirrors the original. You can also download the editable working file if you need to change something downstream.
Getting more from every PDF you translate
The first file is the most expensive. Smartcat stores approved edits, terminology and translation memory, so product names, disclaimers and table headings come out settled in the next PDF, and text you have already translated isn’t paid for twice. Teams translating PDFs as a business process rather than a one-off can read how the AI PDF translation service combines AI agents with expert review, and how online PDF translation scales from one file to a repeatable process.
Security for confidential PDFs
Manuals and catalogs are often pre-release, and contracts are confidential by definition — exactly the files not to feed a free converter. Smartcat is SOC 2 Type II compliant: files are encrypted in transit and at rest, workspaces are isolated per account, and access is role-based.
PDF translation guides, news and help
In-depth how-tos, product updates and documentation on translating PDFs with Smartcat.