Translate MQXLIFF files online. memoQ tags preserved.
Drop your .mqxliff and get it back ready to open in memoQ — translated, with memoQ's curly-tag formatting intact and your locked segments untouched.
What is memoQ MQXLIFF?
MQXLIFF is memoQ's bilingual export format, an XLIFF variant used to move translation jobs between memoQ and external translators or tools. Unlike standard XLIFF, MQXLIFF uses memoQ's proprietary curly-brace tag notation ({1}, {2}, {3>...{<3}) for inline formatting, and carries memoQ-specific metadata: confirmation status, match scores from memoQ's TM, locked segments, and comments.
Translators receive MQXLIFF files from memoQ project managers, work on them in memoQ or compatible CAT tools, and return them with the target column filled. memoQ expects every curly tag in the source to appear in the target with the same number and pairing — any mismatch on import triggers errors that block delivery.
Why memoQ MQXLIFF is tricky for AI translation
- Curly tags are non-XML. Models that handle XLIFF tags fine will tokenize {1} as plain text and reposition or drop it during translation.
- Paired tags ({3> and {<3}) must remain paired and ordered correctly around the same content span. AI models routinely break the pairing.
- Locked segments (mq:locked="true") carry content that must be byte-for-byte identical on return. Any normalization breaks the import.
- Confirmation status must be preserved: segments marked as Reviewer1 or Reviewer2 confirmed need their status carried through, not reset.
- memoQ's match scores (101% in-context, 100% exact, 95–99% fuzzy bands) are stored as metadata. Replacing pre-translated segments destroys the score history.
- Hyperlinks inside curly tags (a common memoQ Word import pattern) need special handling — the link target is not part of the translatable text but the link display text is.
How Fily handles memoQ MQXLIFF
- MemoqCurlyCodec: a dedicated extractor that recognizes {N} and {N>...{<N} patterns, replaces them with indexed placeholders, sends the cleaned text to translation, and reinserts the curly tags in their semantic positions in the target.
- Paired-tag validation: after translation, pairs are verified — every {N> has a matching {<N}, content between them is non-empty, ordering matches the source. Misalignments trigger automatic repair before delivery.
- Locked segment pass-through: segments with mq:locked="true" are copied byte-for-byte to the target. Never sent to the AI.
- Match score preservation: memoQ match-score metadata is carried through unchanged. 100/101 segments are never re-translated.
- Hyperlink-aware extraction: curly tags containing w:hyperlink references are detected; the link target stays bound, the display text is translated independently.
- 12-step QA: glossary enforcement, forbidden-term replacement, tag count repair, no-source-bleed check, and 8 other passes.
Pipeline: mqxliff_memoq_v2@1.0.0
The memoQ MQXLIFF workflow with Fily
Upload
Drop your .mqxliff (single or batch ZIP). Optional: glossary, TM, style guide.
Process
Fily runs the memoQ MQXLIFF pipeline + 12 QA steps. Typical job: 10–20 minutes.
Download
Same format, ready to deliver. QA report HTML attached.
LSPs receive a batch of MQXLIFF files from a client's memoQ project, run them through Fily for first-pass translation overnight, and assign the pre-translated files to reviewers in the morning. Reviewers work in memoQ as usual — tags intact, locked segments untouched, match scores preserved.
Frequently asked about memoQ MQXLIFF
Other formats Fily translates
Ready to translate a memoQ MQXLIFF file?
No card. No setup. Upload one file and see the output.