Merge TMX files into one translation memory
Years of project TMs, one per client or per vendor, each with its own copy of the same sentences. Merging them is easy; merging them without tripling every duplicate and without mixing en-US with en is the part CAT tools leave to you.
━━ the problem
Each project TM carries its own copy of the boilerplate. Importing them one after another into one TM keeps every copy, or overwrites the newer translation with an older one, depending on your import settings.
One export says en-US, the next en_US, a third just en. Merged as-is, some CAT tools treat them as different languages and only part of the TM ever produces matches.
A single file with an invalid character or a cut-off ending makes the merge fail halfway, and you're left guessing which of twenty exports it was.
━━ how TM Cleaner handles it
Drop all the TMX files at once, or a zip of them. Each one is repaired on its own, then they're merged into one TM and duplicates are found across all of them, with your choice of which translation wins. The preview lists every file and how many units it brought.
- ✓up to 20 files per merge, or a zip with up to 50 TMX files inside
- ✓duplicates removed across files, keeping the most common, the newest or your trusted translator's version
- ✓language codes written the standard way with one switch, so every unit lands in the same TM
- ✓a damaged file is repaired, not dropped; its unreadable units are listed
━━ frequently asked
which header does the merged TMX keep?+
the first file's. units keep their own language codes; turn on “fix language codes” to write them all the standard way.
can I merge TMs with different language pairs?+
yes, the result is a multilingual TMX. to import it into Trados or memoQ, keep one pair with “keep only these target languages” in the import check.
is there a size limit?+
50 MB in total without an account, 2 GB with a free sign-in.
━━ related pages
free · no account up to 50 MB · files up to 2 GB · try the demo