Translation
Translation memory
From the Transept teamTranslation memory records the segments you’ve already translated so you never translate the same sentence twice. In Transept it goes one step further than a traditional CAT tool: your past translations are handed to the model as reference on a run, so it reuses your established wording rather than reinventing it each time.
It’s free on every plan, including Free — building and reusing your memory never spends words; only translating new text does.
Added
- A Memory panel in the editor sidebar surfaces past translations for the block you’re on, so you can drop your established wording straight in as a new version.
- Memory as AI reference. Switch memory on for a run — the Memory button on the block toolbar or in the batch dialog — and your matching past translations are fed to the model so it reuses your phrasing. It’s opt-in per run in the editor, so a one-off translation only uses it when you ask.
- A Translation memory settings page under the Library menu sets the defaults every document inherits — whether memory is on by default, what gets searched, and what appears in results (which document statuses, which projects) — with per-document overrides from the editor’s Memory panel.
- Team policy. A team owner can set, and optionally enforce, a memory policy across the team’s documents.
- Import existing memory. Bring a TMX or XLIFF file straight into your translation memory.
Changed
- Workflow steps and Find blocks each carry their own memory setting — use the document default, or force it on or off for that step.
Have a request or hit a snag? Ask Literess in the app, or write to [email protected].