Clean Trados TMX files
Trados Studio TMs accumulate years of debt: every project import, every alignment, every batch task leaves duplicate segments and orphaned inline tags behind. Export to TMX and clean it in seconds — no round-trip through the Studio editor.
━━ the problem
Trados keeps every variant of a segment ever confirmed. Eight near-identical translations of one source all compete for the 100% slot, and the pre-translate step often picks the wrong one — so fuzzy rates quietly drop over time.
Studio's inline tags (bpt/ept/ph) survive TMX export as markup fragments. After a memoQ ↔ Trados hop they turn into stringified junk in the segment body — the engine flags these as suspect markup so you can strip or keep them.
Neural-MT batch pre-translation writes TUs stamped with an MT creationid. Left in the TM they poison your fuzzy matches and any future MT training set. Filter them out by origin in one toggle.
━━ how TM Cleaner handles it
Upload the exported .tmx, preview exactly what would be removed (duplicates, junk, MT-origin units) with the winner of each duplicate group shown, then download a cleaned .tmx plus a re-importable removed.tmx and an audit CSV.
- ✓keeps the best variant per duplicate group — by most-recent, most-popular, or your ranked trusted translators
- ✓strips URLs, numbers, emails, orphan tags, and repeated-punctuation noise that inflates Trados word counts
- ✓cleaned.tmx re-imports straight back into Studio; removed.tmx lets you verify nothing important was cut
- ✓streams multi-GB TMs — a 1 GB Trados export cleans in well under 10 minutes without loading it into RAM
━━ frequently asked
does the cleaned file re-import into Trados Studio?+
yes. the output is a standard TMX 1.4 file. import it as a new TM or into your existing one exactly as you would any TMX. the removed.tmx is also valid TMX so you can inspect (or re-add) anything that was cut.
can I keep my senior in-house translator's version when duplicates conflict?+
yes — priority-author ranking lets you list the changeids of translators whose variants should win. it overrides the popular/latest strategy for any duplicate group where one of them has a variant. no major CAT tool does this at cleanup time.
do you upload my TM anywhere permanent?+
no. the upload goes to storage via a one-time signed URL, is parsed defusedxml-first to neutralise XXE, and is hard-deleted the moment cleaning finishes. download links expire 24h later.
━━ also clean
free tier · pay-as-you-go after · no subscription · see pricing