Translation Memory

Translation Memory (TM) stores every approved translation so the engine can reuse them in future runs. This means consistent translations across your project and lower credit consumption.

How It Works

When Babelize translates a string:

1. The engine checks the TM for an existing translation of the same source string
2. If found, the stored translation is used directly — no AI call needed
3. If not found, the AI translates it and the result is saved to TM for next time

Over time, your TM builds a custom translation database that reflects your team's approved terminology.

Managing TM

Dashboard

1. Open your project
2. Go to the Translation Memory tab
3. Browse, search, and filter by language
4. Edit entries inline or delete outdated ones

CLI

babelize tm list <projectId>             # List all TM entries
babelize tm update <projectId> <id> --translation "new text"
babelize tm delete <projectId> <id>      # Delete a single entry
babelize tm clear <projectId>            # Clear all TM (with confirmation)

TMX Export

Export your TM in the standard TMX format for backup or migration:

# Via API
curl -X GET https://api.babelize.co/api/projects/<id>/tm/export

The TMX file can be imported into most CAT tools or other TM systems.

Plan Availability

Translation Memory is available on all plans, including Free.

Best Practices

• Review translations before they enter TM — use the Review Dashboard to approve translations
• Clean up stale entries — periodically review and remove outdated translations
• Export before major changes — keep a backup before clearing TM