Translation QA Validator
Upload source & translated DOCX - get your file back with QA findings as Word comments.
Glossary (optional)
Add a client terminology list - the validator flags segments where the source term is present but the prescribed translation is missing. For Russian/Ukrainian targets, matching is morphology-aware.
Anonymous limit: 1 MB per file. Sign in for unlimited uploads.
Glossary (optional)
Add a client terminology list - the validator flags segments where the source term is present but the prescribed translation is missing. For Russian/Ukrainian targets, matching is morphology-aware.
Anonymous limit: 4 MB per file. XLIFF/SDLXLIFF declare their language pair; for TMX you can override below.
Glossary (optional)
Add a client terminology list - the validator flags segments where the source term is present but the prescribed translation is missing. For Russian/Ukrainian targets, matching is morphology-aware.
Results
Help us improve
Found a false positive? Missed a real error? Let us know - we read every message.
Questions
Is the QA validator free?
Yes. All 32 surface checks run free and with no sign-up. Only the optional AI semantic check needs a free account, because it calls a language model.
How many languages does it cover?
101 - the full CLDR locale set. Punctuation, quotation marks, digit grouping and spacing rules are applied per locale, not one-size-fits-all.
What does it check?
Placeholder and tag mismatches, number and date changes, punctuation and spacing, doubled words, untranslated segments, glossary and forbidden-term violations, plus your own regex rules - 34 checks in total.
What is the AI semantic check?
An opt-in language-model pass that catches meaning errors the surface checks can't - mistranslations, negation flips, swapped entities. It needs a signed-in account and runs on top of the free checks.
Do you store my text?
No. Segments are validated in memory and the result is returned - nothing is stored on our side.
Is it MQM-compatible?
Yes. Every finding maps to an MQM category and severity, so the report drops into an MQM-based QA workflow.