Translate a Scanned PDF in memoQ: Step by Step

A client sends a six-page scanned contract and a phone photo of a birth certificate, the deadline is Thursday, and your workflow lives in memoQ. You import the PDF and there is nothing to translate. So how do you translate a scanned PDF in memoQ? memoQ 12.5 can’t read the page for you, so the job splits into three parts: OCR outside memoQ, a clean import, and a delivery format the client accepts.

Below is the hands-on version for memoQ 12.5 (desktop; memoQ TMS uses the same help pages): the import path, the filter settings, the Word cleanup, text inside stamps and logos, and a pre-import checklist. For the background on why CAT tools fail on image-only PDFs, read why Trados and memoQ can’t import a scanned PDF first.

What memoQ 12.5 does with a scanned PDF or a photo¶

memoQ 12.5 has no built-in OCR (optical character recognition, the step that turns letters in a picture into text). The current version is memoQ 12.5.20, with release notes dated 3 September 2026 on the memoQ downloads page, and its PDF help page is explicit about scans:

“Can’t import scanned PDF files: memoQ doesn’t extract text from scanned PDF files, where the pages are saved as images and not as text.”

The same page sends you to an external page-reader program such as “Nuance OmniPage or ABBYY FineReader”. memoQ’s own position is OCR outside, translation inside.

memoQ did have OCR once. memoQ 8.1 (2017) integrated Iceni’s TransPDF service, and the memoQ 8.1 announcement priced it in credits: “Each credit allows you to OCR one page and will cost you about 15 euro cents”. The same post, by Peter Reynolds of memoQ on 26 May 2017, opens with a line most translators will recognize:

“Most translators I have met passionately hate image-based PDFs”

The feeling outlived the feature. The memoQ release notes for 10.4.12 (8 November 2023) list “Removing the TransPDF import option … Because of the discontinuation of the TransPDF service”. So paid OCR inside memoQ existed from 8.1 to 10.3 and has been gone since 10.4.

Nothing replaced it. The release notes from 10.4.12 through 12.5.20 contain no PDF or OCR feature; memoQ’s AI work in that period went into AGT translation and AIQE quality estimation, which handle text that already exists.

For PDFs that do contain text, the PDF help page lists two import methods (memoQ 12.5):

Conversion to DOCX. memoQ converts the PDF to Word with the Aspose.PDF library, in “Text flow conversion” or “Attempt to keep formatting” mode, and imports that.
Plain text. memoQ pulls the raw text with xPDF and discards the formatting; the help reserves it for LiveDocs corpora and alignment, not documents you translate.

Photos are a separate trap. The supported formats list does include “Image files” (.bmp, .jpg, .png, .tif and others), but memoQ’s help has no page explaining that filter, and nothing in the help says memoQ reads text from a photo. Treat a phone photo exactly like a scan: OCR first.

Not sure whether your PDF is a scan? Try to select a sentence in a PDF reader. If nothing highlights, or the paste is gibberish, it’s an image-only file; the guide to converting a scanned PDF to editable Word covers that test in detail.

Why a scan breaks in a CAT tool: the mechanism¶

A scanned page is a photograph of paper. memoQ doesn’t look at pages: it reads text, splits it into segments (the sentences you translate one by one), and keeps the formatting around them as tags. A scan has no text to split, so memoQ has nothing to work with until something reads the picture.

OCR does two separate jobs. First it recognizes characters: which marks are letters, in which language, in which order. Then it rebuilds a document: which lines form a paragraph, whether a grid is a table, whether the page has two columns. A tool can do the first job well and the second badly, and you get correct words in the wrong order.

The hardest parts of a real document aren’t plain text. Stamps, seals, signatures, handwritten dates and QR codes are pictures inside the picture: OCR skips them, reads them as garbage, or drops part of a seal’s ring text into a paragraph. Tables and columns add a reading-order problem, explained in why CAT tools break PDF tables and multi-column layouts.

Rebuilding a Word copy of a scan is the job we built ChatsControl around, so here is the full job from the inside. A vision model looks at each page and builds its Word copy: text typed as printed, tables as real tables, fields next to their labels. Stamps, seals, signatures and photos are cut from the page and placed where they stand; QR codes are read and drawn again clean. A second check compares the copy with the page line by line and sends what it finds back to be fixed. Each scan gets a quality grade (good, fair or poor); handwriting and very degraded scans are read unreliably, and the grade says so.

For memoQ users the point is simple: OCR quality and layout quality are two different questions, and memoQ can’t fix either after import. The text order and structure you hand memoQ become your segments.

Route 1: OCR the scan into a searchable PDF, then use memoQ’s PDF filter¶

Route 1 keeps the PDF as the file you import. An OCR program adds a text layer to the scan (a “searchable PDF”; Adobe’s Acrobat help describes this step), and memoQ’s PDF filter converts that layer to DOCX with Aspose.PDF. Import it with options so you see the filter settings; the path below comes from memoQ’s document import options and supported formats help (memoQ 12.5):

Open the project and go to Project home > Translations.
On the Documents ribbon, click the arrow under Import and choose Import with options.
Select the searchable PDF and open it.
In the Document import options window, select the file and click Change filter and configuration.
In Document import settings, make sure Filter is set to PDF (Portable Document Format).
Choose conversion to DOCX, not plain-text import (xPDF), which throws away formatting.
Pick the mode: Text flow conversion (the label warns it “might slightly change formatting”) or Attempt to keep formatting (the label warns that some text may be lost).
Check the other options: Specify relative horizontal proximity, Recognize bullet point symbols, and the DOCX options tab.
Finish the import and compare the document with the scan.

For a scan, start with Text flow conversion: a formatting difference is visible and fixable, while a missing line is an omission nobody notices until the client does. Use Attempt to keep formatting only for a layout-heavy page you have time to check line by line.

Route 1 has a structural limit. The searchable PDF goes through two reconstructions you don’t control: your OCR program builds a text layer, then Aspose.PDF guesses paragraphs and tables from it. You can’t fix a broken table or a misread name in between, so every mistake becomes a segment. Use Route 1 for a clean, single-column letter or a short certificate with no tables; for grids, forms or two columns, Route 2 gives you more control.

Route 2: OCR to DOCX first, clean it, then import it as a Word file¶

Route 2 is the route we recommend for most scans: OCR the page to a Word document outside memoQ, clean it with the scan open next to it, and import the result like any client .docx. You fix problems in Word, where you can see them, instead of meeting them as broken segments.

The advice isn’t new. In a 2012 post about the memoQ PDF filter (memoQ 5/6 era, so read it as history), Kevin Lossner put it in one line:

“A PDF with complex or scanned content requires tools such as OCR software (OmniPage, ABBYY FineReader, etc.)”

Fourteen years later, memoQ’s current help says the same. What changed is the choice of OCR tools.

Which OCR tool to use for the DOCX¶

ABBYY FineReader or Nuance OmniPage. Dedicated OCR programs, the two memoQ’s help names.
Microsoft Word. Word can open a PDF and convert a copy to DOCX, but Microsoft says the conversion “works best with files that are mostly text”. A page that is mostly a picture may come through as a picture.
Adobe Acrobat. Recognizes text and creates a searchable PDF, which you can then save as Word or feed to Route 1.
Google Docs via Google Drive. Free for a simple page, with limits covered below.

Google Drive’s OCR help sets a 2 MB limit for this method (as of October 2026) and is candid about structure:

“Lists, tables, columns, footnotes, and endnotes are not likely to be detected.”

So Google Docs is fine for a one-column letter and risky for a form or a bank statement. Run one sample page through your tool before committing a 40-page job to it.

Cleanup before import: what to fix in the DOCX¶

Do these with the scan open side by side; each fix removes a problem you’d otherwise meet in memoQ’s translation grid.

Check names, numbers and dates against the scan. OCR confuses look-alike characters, and a wrong digit in a passport number survives translation. Mark what you can’t read as [illegible].
Remove line breaks inside sentences. OCR can end every printed line with a paragraph mark. memoQ segments within paragraphs, so a sentence cut into three lines becomes three segments that make no sense alone.
Pull text out of text boxes and frames. OCR tools sometimes put each block in its own box to mimic the page, and boxes can reach memoQ out of reading order. Move the text into the body.
Rebuild tables as real Word tables. If a table came out as tabs and spaces, its cells land in one paragraph and memoQ treats them as running text. Insert a proper table and move the text into it.
Clear noisy formatting. Font, size or character spacing that changes mid-sentence shows up in memoQ as extra tags. Clear direct formatting in body text, then re-apply bold where the original has it.
Decide what to do with stamps, seals and signatures. If the receiving authority wants the seal’s wording, add a bracketed note such as [Seal: …] in the text. The guide to translating stamps and seals covers the wording conventions.
Save as .docx and keep the scan for reference during translation.

Then import the DOCX through Import or Import with options like any other Word file, and scroll the translation grid and the tag view against the scan once; automated QA checks for tags, numbers and terminology catch what the eye misses later. For a very faint or blurred page, fix the source first: what to fix in a poor-quality scan.

Route 2 costs time: on a dense form the cleanup can take longer than the translation. Quote it as a separate line, as shown in how to quote a scanned document when word count finds nothing.

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

Route 3 swaps the OCR program and most of the cleanup for a service. 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 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 memoQ like any client Word file (or have it translated directly in the same service).

The CAT route works like this:

Upload the scan or photo to ChatsControl and wait for the job to finish.
On the job page, download the Word copy of the original (the Original button, next to the translated Word file and the raw scan).
Check that copy against the scan: names, numbers, dates, and the quality grade.
Import the source-language .docx into memoQ and translate with your TM and termbase.

The result is an ordinary Word document; memoQ imports it the way it imports any client .docx. Check the tag view after import, as with any converted file, and see real scans next to the Word files the service returned before you upload anything.

Prices as of October 2026, per the ChatsControl app: 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 recognized characters, with a minimum of 1,000 characters per page. The job also produces a translation, and you pay for the upload as above; there’s no separate OCR-only price. Uploads are deleted automatically after 30 days by default.

The limits:

ChatsControl is not a CAT tool or a TMS: no translation memory, no segment-level project management.
It’s cloud only, with no on-premise or offline mode, so it doesn’t fit clients who forbid cloud processing.
Handwriting and very degraded scans are read unreliably, and the quality grade says so (more on handwritten documents and OCR).
Stamps, seals and signatures stay as pictures where they stand. If the authority wants the seal’s text rendered, you add a bracketed note such as [Seal: …] yourself; the stamp’s text is not translated inside the picture.

After translation in memoQ, you can run the exported file against the source through the free translation QA check, which compares numbers and meaning line by line.

The three routes compared¶

No route makes the checking step optional; the routes differ in where you do it.

	Route 1: searchable PDF + memoQ PDF filter	Route 2: OCR to DOCX + cleanup	Route 3: scan-to-Word service
Who reads the scan	Your OCR program	Your OCR program	The service
What you import into memoQ	The searchable PDF	Your cleaned .docx	The service’s .docx of the original
Who rebuilds the layout	Aspose.PDF inside memoQ, no edit step	You, in Word, before import	The service; you check it
Can you fix errors before segments exist	No	Yes	Yes
Export from memoQ	DOCX (TXT for plain-text import)	DOCX	DOCX
Cost	Your OCR license	Your OCR license or a free tool, plus cleanup time	Free tier 6,000 characters/month, then $1.50 per 1,000 characters (Oct 2026)
Works offline	Depends on the OCR program	Depends on the OCR program	No, cloud only
Suits	Clean single-column letters	Forms, tables, mixed layouts when you have time	Stamps, tables, photos of documents, tight deadlines

What survives each route is the other half of the decision:

Element on the scan	Route 1	Route 2	Route 3
Printed text	Yes, if the OCR text layer is right	Yes, after your check	Yes, typed as printed; check it
Tables	Guessed by Aspose.PDF from the text layer	Depends on the OCR tool; rebuild by hand if lost	Rebuilt as Word tables
Stamps, seals, signatures	Text in images not recognized	Depends on the OCR tool	Cut from the page, placed where they stand
QR codes	Not read; at best stays a picture	Depends on the OCR tool; check it	Read and redrawn clean
Handwriting	Check word by word	Check word by word	Read unreliably; the quality grade flags it
Reading order of columns	Set by Aspose.PDF, no edit step	You fix it before import	Follows the page; check it

What goes wrong with scans in memoQ, and what you deliver¶

Each trap below traces back to memoQ’s own documentation or release history.

Tutorials that still point to TransPDF¶

The memoQ 8.1 blog post from 2017 is still online, with its “Pay for OCR” button and 15-cent credits. memoQ’s PDF help now says “The TransPDF service is no longer available in memoQ 10.4 and later.” An internal SOP or training video that mentions TransPDF predates memoQ 10.4 and needs rewriting.

“Attempt to keep formatting” can drop text¶

memoQ labels its second DOCX conversion mode with its own warning that some text may be lost, per the PDF help page. On a scan with stamps, small print and boxed fields, a lost line is an omission in a legal document. If you use this mode, count the paragraphs on the scan and in the memoQ document before you translate.

Plain-text import strips the structure¶

On a form, the xPDF plain-text import loses which value belongs to which label, so a date of birth and a date of issue become two loose dates.

Text in images stays invisible¶

Even a PDF with a perfect text layer can hide text inside pictures: a logo, a pasted-in stamp, a signature block. memoQ’s PDF help says it directly:

“if the PDF document contains text in images, memoQ will not recognize the text in there.”

memoQ’s answer is Image transcriptions, and it’s manual: you “type the text that you see in the image”, translate those transcriptions in memoQ, and send an image localization pack to a graphic artist who puts the translated text back into the picture. It works only for images embedded in documents you import with the “Import embedded images” option selected. For a stamp on a certificate, a bracketed note in the text such as [Seal: …] is the faster answer than editing the picture.

Password-protected PDFs don’t import¶

memoQ’s PDF help says password-protected PDFs can’t be imported. Ask the client for an unlocked copy before you quote; it isn’t an OCR problem and no OCR setting will fix it.

Delivery: memoQ never exports a PDF¶

memoQ exports a PDF source as DOCX or TXT, depending on the import method, never as PDF (PDF help, memoQ 12.5), and Routes 2 and 3 give you a DOCX too. To deliver a PDF, check the exported DOCX against the scan and save it as PDF from Word. If the client expects a page-identical PDF, that is layout work; agree on it and price it before you start (how agencies charge for DTP on scanned documents).

The memoQ PDF preview tool doesn’t change this: memoQ describes it as a separate application that shows a reference PDF next to the editor. A viewer, not a converter.

Pre-import checklist for a scan in memoQ¶

Run this list before you create the memoQ project; every row is cheaper to fix now than after translation.

#	Check	How	Why it matters in memoQ
1	Does the PDF have a text layer?	Try to select and copy a sentence in a PDF reader	memoQ doesn’t extract text from scans; no text layer means OCR first
2	Is the PDF password-protected?	Open it; look for a password prompt	memoQ can’t import password-protected PDFs
3	Is there a native file?	Ask the client for the Word original	A native .docx skips OCR entirely
4	Which route?	Clean letter: Route 1; tables or forms: Route 2 or 3	Route 1 has no edit step
5	Is the OCR language right?	Set the source language in the OCR tool	Wrong language means wrong characters, which become wrong segments
6	Names, numbers, dates	Compare with the scan character by character	A misread digit in the source survives translation
7	Line breaks inside sentences	Show paragraph marks in Word	Each break splits a segment
8	Text boxes and tables	Move boxed text into the body; rebuild tables	Boxes come in out of order; fake tables read as running text
9	Formatting noise	Clear direct formatting in body text	Font and spacing changes become extra tags
10	Stamps, seals, signatures, logos	Add [Seal: …] notes or plan Image transcriptions	memoQ doesn’t recognize text in images
11	Import settings for a PDF	PDF filter, DOCX conversion, Text flow first	Plain text strips formatting; keep-formatting mode may lose text
12	Delivery format	Agree DOCX or PDF with the client before you start	memoQ exports DOCX or TXT, never PDF

If the files arrive as phone photos through a messenger, first get every page in focus, flat and in order; the intake workflow for WhatsApp document photos has a ready-made version. For the same job in other tools, see the Trados Studio version of this guide and the Smartcat one.

FAQ¶

How do I translate a scanned PDF in memoQ?¶

Run the scan through OCR outside memoQ, check the result against the page, and import the cleaned DOCX like any Word file. If you only have a searchable PDF, import it with Import with options, the PDF filter and DOCX conversion in Text flow mode.

Does memoQ have OCR?¶

No. memoQ 12.5 doesn’t extract text from scanned PDFs and recommends an external program such as ABBYY FineReader or Nuance OmniPage. The paid TransPDF OCR that memoQ 8.1 offered in 2017 was removed in memoQ 10.4.

Which memoQ PDF import settings should I use for an OCR’d PDF?¶

Use the PDF filter with conversion to DOCX, not plain-text import, which discards formatting. Start with Text flow conversion, because the other mode, Attempt to keep formatting, warns that some text may be lost.

Can memoQ translate text in images?¶

Not automatically. memoQ’s help says text in images inside a PDF isn’t recognized. For images embedded in an imported document, the Image transcriptions feature lets you type the visible text by hand, translate it and send an image localization pack to a graphic artist.

Will the segments and tags be messy after OCR?¶

OCR output can make them messy: line breaks inside sentences split segments, and font or spacing changes inside a sentence turn into extra tags. Clean the DOCX before import and check the tag view after import, as with any converted file.

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

Check the client’s confidentiality terms and the service’s retention rules first. Installed OCR software keeps the file on your machine; a cloud service doesn’t, so pick one with clear deletion settings or keep the job offline.

Can memoQ give me a translated PDF back?¶

No. memoQ exports a PDF source as DOCX or as plain text, depending on how you imported it. If the client needs a PDF, save the checked Word file as PDF and budget for the layout work in between.

Sources¶

Try ChatsControl

AI platform for professional translators

Try for free →