How to Translate a Scanned PDF in Trados Studio 2024 and 2026

A client sends a scanned marriage certificate as a PDF, you add it to a Trados Studio 2024 project, and Studio either refuses the file or reports zero words. Here’s the short answer to how to translate a scanned PDF in Trados Studio: current Studio has no OCR of its own, so the scan has to be read somewhere else first. Below are the three routes that work in Studio 2024 and the 2026 Release, the checks after import and what you can deliver back.

The “why” behind failed imports is covered in our explainer on scanned PDF import failures in Trados and memoQ; this guide is the hands-on part.

What Trados Studio 2024 and 2026 actually do with a scanned PDF¶

Trados Studio has had no built-in OCR since the PDF conversion change RWS announced in March 2023. According to the announcement by Daniel Brockmann of RWS (3 March 2023), the cloud platform switched to a new PDF conversion technology on 13 March 2023, and Studio desktop uses it starting from cumulative update 6 of Studio 2022. Every later release, Studio 2024 and the 2026 Release included, inherits it.

The same announcement spells out what that means for a scan:

“If a PDF file contains merely a scanned picture … the new technology will not be able to convert the document.”

The key word is “merely”. If the PDF already carries a selectable text layer (someone ran OCR on it before sending), Studio still attempts the conversion. If the page is only a picture, no project setting helps. Not sure which kind you have? Use the select-and-copy test from the import-failure explainer.

Paul Filkin of RWS put it even more bluntly in an RWS Community thread about the IRIS plugin:

“the new filetype doesn’t OCR at all”

Several PDF settings disappeared in the same change. Per the 2023 announcement, “Headers and footers” and “Detect tables” are no longer settings (both are always extracted), “Image recovery” is gone (images are kept, but no text is extracted from them) and “Recognize PDF text” is gone. One new option arrived: “Use alternative processing (better for non-Latin based languages)”.

The 2026 Release didn’t bring OCR back. The Trados Studio 2026 release notes (the release launched on 2 July 2026) cover the 64-bit build, .ttb termbases, AI Bridge and Language Weaver Pro, and say nothing about PDF or OCR. The Studio 2026 page on configuring the PDF file type lists three layout recover modes and the alternative processing switch, with no OCR option anywhere.

Version timeline: which advice fits your Studio¶

The versions above are as of October 2026, taken from the 2023 announcement, the PDF Assistant listing in the RWS AppStore and the 2026 release notes.

The 2019 tutorial in the first row is Emma Goldsmith’s “How to translate a PDF in Trados Studio” on the Trados blog (1 May 2019). RWS Community member Daniel Hug summed up its status in a thread titled “PDF not converted into Word”: it’s “accurate for Trados Studio 2019-2022, but I think a lot of 2024 users are surprised”. The IRIS route is closed as well, since the IRIS PDF OCR Support listing now says “This is no longer being maintained” and no Studio file type has supported IRIS since March 2023.

The surprise hits people who upgraded for other reasons. In the “Trados Studio 2024 does not open PDF files” thread, a translator who moved from Studio 2019 to 2024 wrote:

“I wouldn’t have bought the update if I had known of this issue beforehand”

In the same thread Paul Filkin explained that the OCR vendor “made it impossible for us to work with them any more”. So the missing OCR isn’t a bug waiting for a patch, and nothing RWS has published as of October 2026 brings it back.

Image files and phone photos¶

Studio 2024’s list of supported file types includes *.pdf but no JPG, PNG, TIFF or BMP. A phone photo of a passport or a TIFF from an office scanner can’t go into a project at all. Turn it into a DOCX with OCR first; the intake side is covered in our guide to document photos sent over WhatsApp.

Trados Cloud¶

Trados Cloud (Trados Team, Accelerate and Enterprise) moved to the same PDF technology on 13 March 2023, per the RWS announcement. The current Trados Cloud “PDF settings” help page still lists “Image recovery” and “Recognize PDF text”, the options the 2023 announcement says were removed, so treat that page as stale. PDF Assistant is desktop only, so for a cloud project do the OCR before upload and send the DOCX.

Why a scan breaks in a CAT tool¶

A scanned PDF is a photograph of a page wrapped in a PDF container. Trados Studio works on text: it splits sentences into segments, matches them against your translation memory and counts words. A photograph has no sentences, so something has to read the page and type out what’s printed first. That step is OCR (optical character recognition).

Reading the text is only half the job. A Word file also needs structure: which lines form one paragraph, which boxes are table cells, where a column ends. A PDF stores only characters placed at coordinates, so every converter guesses the structure back. When the guess is wrong, Studio gets a sentence split across two segments or a table read across columns. Our breakdown of why CAT tools break PDF tables shows what that looks like in the Editor.

Official documents add a third problem. Stamps, seals, signatures, handwritten dates and QR codes aren’t typed text, so OCR skips them or turns the ring of text around a seal into nonsense inside a paragraph. Converters also switch fonts or character spacing mid-line where the scan was uneven, and Studio shows each switch as a tag.

ChatsControl’s scan reader treats layout as its own job. A vision model looks at each page and rebuilds it as a Word copy, the copy is rendered, and a second check compares it with the page line by line before the scan gets a quality grade (good, fair or poor). Handwriting and very degraded scans still read unreliably, and the grade says so. The scan examples page shows real scans next to the Word files that came back.

Route 1: PDF Assistant for Trados, step by step¶

PDF Assistant for Trados is a free RWS AppStore plugin that converts a PDF into a DOCX before translation through the Microsoft Word desktop API. Word does the conversion (and the OCR, for the images you select), so Microsoft Office has to be installed on the same machine. As of October 2026, the AppStore listing shows these versions:

• 3.0.2.1 (16 September 2026) for Trados Studio 2026 Release;
• 2.1.0.0 (22 May 2025) for Studio 2024 SR1;
• 2.0.1.0 (13 June 2024) for Studio 2024;
• 1.1.1.0 for Studio 2022 SR1 and 1.0.1.1 for Studio 2022.

The 3.0.2.1 changelog lists “Fix OCR failure”, a fix for hung conversions, and support for pages bigger than Word’s 22 x 22 inch limit.

How to run it¶

1. Check that desktop Microsoft Word is installed on the machine where Studio runs.
2. In Studio, open Add-Ins > RWS AppStore, find PDF Assistant for Trados and install the version that matches your Studio release.
3. Open the plugin from the Add-Ins tab > Toolbox.
4. Add the scanned PDF. In the provider step pick Microsoft Word; the plugin documentation says “For now there is only Microsoft Word to select from”.
5. Tick the embedded images you want OCRed. On a scan the whole page is an image, so the text you need sits inside the images you tick.
6. Run the conversion and save the DOCX.
7. Open the DOCX in Word and compare it with the scan page by page before Studio sees it: names, dates, numbers, table cells, reading order.
8. Add the checked DOCX to your Trados project (or open it as a single document) and translate it like any client Word file.

Where PDF Assistant falls short on scans¶

The plugin’s own documentation is upfront about the limits. It warns that “when you OCR the images you will lose any background image”, and on a certificate the background is where the letterhead, the coat of arms, the stamp and the signature live. It also says the conversion “is not a process that is guaranteed to work 100% of the time” and points to ABBYY FineReader or Acrobat Pro for hard cases.

Real scans show what that means in practice. Under the 2023 RWS announcement, Mafalda Gouveia da Costa described PDF Assistant’s output on scanned PDFs this way:

“approx. 70% of text processed and 30% of chunks of images with text… (a DTP nightmare)”

Text left inside an image never reaches a segment, so it’s missing from the word count and the translation unless someone retypes it. PDF Assistant suits clean, typed scans where the page background doesn’t matter.

Route 2: OCR to DOCX first, then import into Studio¶

RWS’s own recommendation is to do the OCR outside Studio. The March 2023 announcement lists opening the PDF (including an OCRed PDF) in Microsoft Word and saving it as DOCX, Adobe’s PDF to Word conversion, and third-party OCR such as ABBYY FineReader or Readiris. Our comparison of Word, Google Drive, Acrobat and ABBYY for scans covers each tool’s limits.

Paul Filkin described his own approach in the IRIS thread: run the free Tesseract-based FreeOCR, then “recreate the layout in Word myself”. Even RWS staff rebuild the layout by hand. You can hand the OCR result to Studio in two ways.

Option A: a searchable PDF through Studio’s PDF filter¶

If your OCR tool saves a searchable PDF (the scan image with an invisible text layer on top), Studio’s PDF conversion will attempt it, because the text is now selectable. The Studio 2026 PDF file type settings then decide how the page is rebuilt. You’ll find them under Projects view > Home tab > Project Settings for the active project, or File > Options for all future projects, then File Types > PDF (sub-pages Converter, Common, QuickInsert and Font mapping).

The layout recover mode matters most here:

• Flowing recovers page layout, columns and graphics and preserves text flow.
• Continuous detects layout and columns but only recovers formatting, graphics and text flow.
• Exact recovers the exact page presentation using text boxes in Microsoft Word.

For an OCRed scan, start with Flowing (my recommendation, not an RWS one). Exact gives a target that looks like the page, but each text box is a separate island, and a sentence that wraps across two boxes turns into two segments. For Arabic, Hebrew, Chinese, Japanese or any other non-Latin source, also tick “Use alternative processing”, which RWS describes as better for non-Latin based languages.

Option A skips cleanup, so it’s fast and risky: a misread digit can only be fixed in the target.

Option B: a DOCX you clean up first¶

Option B takes longer and gives you control. Convert the scan to DOCX, then fix it in Word before import:

1. Line breaks inside sentences. OCR puts a hard return at the end of every printed line. Studio treats each one as a segment boundary, so join the lines back into paragraphs.
2. Text boxes and frames. Layout-faithful OCR output wraps text in boxes. Move body text into normal paragraphs, because Studio segments boxes separately.
3. Reading order. Check that two-column pages and tables run the way a person reads them.
4. Font and spacing noise. Clear stray character spacing, scaling and font switches from the body text. Each one becomes a tag in Studio.
5. Numbers and names. Check every date, ID number, amount and proper name against the scan.
6. Unclear spots. Where the scan is unreadable, type a marker such as [illegible] instead of guessing.

Then add the DOCX to the project as you would any client Word file. If the scan itself is faint or skewed, fix the image before OCR, not after.

When neither option produces something usable, translators fall back to typing. In the IRIS thread, Antonella Pecchioli spent a day fighting a scanned certificate and wrote:

“In the end, I wrote the document by hand in a word file”

Retyping is fair for a one-page certificate if you price it in; our breakdown of what retyping scans really costs shows how fast the hours add up.

Route 3: a scan-to-Word service, then Trados Studio¶

One option is ChatsControl: you upload the scan or photo (PDF, PNG, JPG, TIFF, WebP), it reads the page and returns an editable Word copy that follows the page layout. Text is typed as printed, tables come back as real tables, stamps, seals, signatures and photos are cut from the page and placed where they stand, and QR codes are redrawn clean. You then import that .docx into Trados Studio like any client Word file (or have it translated directly in the same service).

The result is an ordinary Word document, and Trados Studio imports it the way it imports any client .docx. For a CAT project the steps are:

1. Upload the scan or photo and wait for the job to finish.
2. On the job page, download the Word copy of the original (the “Original” button, in the source language). The translated Word file and the raw scan sit next to it.
3. Check the Word copy against the scan as in Route 2: numbers, names, table cells, anything the quality grade flags.
4. Add the .docx to your Trados project and translate it with your own TM and termbase.
5. Check the tag view after import, as with any converted file.

Each upload also produces a translation and is billed per recognised character; there’s no separate OCR-only mode. As of October 2026, ChatsControl has a free tier of 6,000 characters per month, then pay as you go at $1.50 per 1,000 characters ($2.70 per 1,800-character page). Scans are billed by recognised characters but at least 1,000 characters per page, and uploads are auto-deleted after 30 days by default.

ChatsControl isn’t a CAT tool or a TMS: there’s no translation memory and no segment-level project management, which is why you bring the copy back into Studio. The service is cloud only, with no on-premise or offline mode. Handwriting and very degraded scans read unreliably, and the quality grade tells you when that happened.

Stamps, seals and signatures stay as pictures where they stand, and their wording isn’t translated inside the picture. If the receiving authority wants the seal’s text rendered, add a bracketed note such as [Seal: …] to your translation, as our guide to translating stamps and seals describes.

Three routes compared¶

No route wins everywhere. A typed two-page letter on Studio 2026 with Word installed is a fair job for PDF Assistant, while a notarised power of attorney with a seal over the signature line needs a route that keeps the page background.

Colleagues on other tools can see scanned PDFs in memoQ and Smartcat’s built-in OCR; Studio’s own editions are in our Trados Studio pricing breakdown.

What goes wrong after import, and what you deliver¶

The analysis shows 0 words or only the first page¶

In a Trados 2024 thread, Olga Gallegos reported this after running PDF Assistant:

“the file converted shows the count words “0” and only the first page for translation. The source PDF file has more than 200 pages.”

When the count comes back as 0, open the intermediate DOCX in Word before touching any Studio setting. If its pages are pictures rather than typed text, the OCR step didn’t run on them and Studio is counting correctly. Update PDF Assistant (3.0.2.1 fixed an OCR failure and hung conversions, per its AppStore listing) and convert again, or switch to Route 2.

A zero count also breaks your quote. Mafalda Gouveia da Costa described the agency side under the 2023 announcement:

“for an everyday routine in which we receive translation requests in paper (certified translations) we suddenly lost the ability to give a quick word count/quotation to our clients”

Count after OCR and quote the OCR and checking time separately; our guide to quoting a scanned document when the word count finds nothing walks through the numbers.

Wrong PDF settings for an OCRed PDF¶

An OCRed PDF converted with Exact comes into Studio as many text boxes: segments break mid-sentence and the order follows box positions. Switch to Flowing and re-add the file before translating. A searchable Arabic or Chinese PDF that arrives garbled deserves a second conversion with “Use alternative processing” ticked, the option RWS added in the 2023 change. In any mode, open the tag view on the first pages and look for tag clusters around single words; our article on automated QA checks for tags and numbers covers what to verify before delivery.

Text inside images stays untranslated¶

Since 2023 Studio keeps images but extracts no text from them, so a stamp, a seal or a letterhead logo stays in the source language in your target. Add bracketed translator’s notes where the reader needs them.

The client wants a PDF back¶

The target of a PDF source in Trados is a Word file. Alison Field of RWS put it plainly in the same Trados 2024 thread:

“Trados always creates a Word document from PDF files for translation”

Tell the client at intake that the deliverable is a DOCX, or plan the PDF step: fix overflowing text in the target DOCX and save it as PDF from Word. Rebuilding the page look is layout work, and our guide to DTP charges for scanned documents shows how agencies bill it. Before sending, the free QA validator can compare the translated file with the source for numbers and meaning.

Pre-import checklist for a scan in Trados Studio¶

FAQ¶

How do I translate a scanned PDF in Trados Studio 2024 or 2026?¶

Turn the scan into a Word file first with the free PDF Assistant plugin (which uses desktop Word), an OCR tool such as Acrobat or ABBYY FineReader, or a scan-to-Word service. Check the DOCX against the scan, then add it to your Trados project.

Does Trados Studio have built-in OCR?¶

No. Since the PDF conversion change of March 2023 (Studio 2022 CU6 and every later release, Studio 2024 and 2026 included), Studio can’t convert a PDF that is merely a scanned picture. A scan that already has a selectable text layer still converts.

Can Trados Studio translate a JPG, PNG or TIFF image?¶

No. Studio 2024’s list of supported file types includes PDF but no JPG, PNG, TIFF or BMP. Run OCR on the photo or image first and import the resulting DOCX.

Does PDF Assistant for Trados OCR scanned PDFs?¶

Yes, through Microsoft Word: you tick the embedded images to OCR and Word converts them, so desktop Office must be installed. RWS’s docs warn that OCRing images removes any background image and that the conversion isn’t guaranteed to work every time.

Can I deliver a translated PDF back from Trados Studio?¶

Not directly. Trados always creates a Word document from a PDF source, so the target is a DOCX. Save it as PDF from Word after checking the layout, and agree any desktop publishing with the client up front.

Will the segments and tags be messy after OCR?¶

Expect some noise. Line breaks inside sentences split segments, text boxes break reading order, and OCR font or spacing changes show up as tags. Clean the DOCX before import and check the tag view after.

Is it safe to upload client scans to an online OCR service?¶

Check your client agreement first. PDF Assistant and desktop OCR tools keep the file on your machine, while a cloud service stores it for a time; ChatsControl auto-deletes uploads after 30 days by default (as of October 2026).

Sources¶

1. Announcing change to PDF conversion technology in Trados Cloud offerings and Trados Studio
2. Configuring the PDF file type - Trados Studio 2026 Release
3. Trados Studio 2026 Release - release notes
4. PDF Assistant for Trados - RWS AppStore
5. IRIS PDF OCR Support - RWS AppStore
6. Supported file types - Trados Studio 2024
7. PDF settings - Trados Enterprise and Accelerate
8. How to translate a PDF in Trados Studio
9. App IRIS PDF OCR Support doesn’t work
10. Trados 2024 does not open PDF files
11. Trados 2024 - error: object reference not set to an instance of an object
12. PDF not converted into Word