Clean OmegaT translation memories
OmegaT is unusual among CAT tools: TMX is its native memory format, so there's no export step — your project_save.tmx IS the translation memory. That also means it accumulates every edit you ever made, and there's no built-in cleanup. Upload it here and hand it back a leaner TM.
━━ the problem
OmegaT appends a new translation unit each time you change a segment, so a long project's project_save.tmx holds many stale variants of the same source alongside the final one. Nothing in OmegaT collapses them — the file just grows.
Delete or swap a source document and its segments stay in the TM. Over a multi-file project these orphans pile up as duplicates and near-duplicates that dilute your matches without ever appearing in the editor again.
OmegaT's level-2 TMX stores inline formatting tags that shift shape between versions and across imported TMs. Round-tripping through other tools can leave tag-only or markup-heavy segments that are pure noise for matching.
━━ how TM Cleaner handles it
Upload project_save.tmx (or any TMX in your /tm folder), preview the duplicates and junk with the winner of each group shown, and download a cleaned TMX you can drop straight back into the project. It's a standard TMX 1.4 file, so OmegaT reads it with no conversion.
- ✓no export/import dance — OmegaT's memory is already TMX, so you clean the real file
- ✓collapse the accumulated edit-history variants to one authoritative unit per source
- ✓strip tag-only and markup-heavy segments left by cross-tool round-trips
- ✓free tier fits solo translators, NGOs, and volunteer projects that pick OmegaT for exactly that reason
━━ frequently asked
which OmegaT file do I upload?+
project_save.tmx from your project's /omegat folder is the main working TM. you can also clean any TMX in the project's /tm reference folder, or a TM you exported to share. all are standard TMX and clean the same way.
will the cleaned TM still load in OmegaT?+
yes — the output is TMX 1.4, exactly what OmegaT reads. drop it back in place of the original (keep a backup, or use the removed.tmx to verify nothing important was cut before you replace the working file).
does OmegaT have its own cleanup for this?+
not really — OmegaT has no built-in dedup or junk removal for the working TM, which is why the file grows unbounded. this fills that gap without adding a plugin or leaving your open toolchain.
━━ also clean
free tier · pay-as-you-go after · no subscription · see pricing