Skip to main content
Native workspace translation memories support interchange through the Cloud UI and authenticated API. External TMS memories follow different rules when they are not mirrored locally.

​
Supported formats

TMX is TMX 1.4 with inline codes, tuid upsert, and locale-pair variants. CSV uses a Hyperlocalise row layout (header row with source_locale, target_locale, source_text, target_text, match_score). The importer also accepts Crowdin-style locale-matrix CSV (locale codes as column headers).

​
Import in Cloud

  1. Open Translation Memories and select a workspace memory you can edit.
  2. Choose Import and upload CSV or TMX.
  3. Review the import preview, then confirm. Large files are bounded by documented size and unit limits; oversize imports fail with an explicit error.

​
Export in Cloud

  1. Open the memory and choose Export.
  2. Select TMX or CSV.
  3. Either download all locales or pick a source and target locale pair to filter the file.

​
API

Export:
Import (preview with dryRun: true):
Body includes format (csv or tmx), content, and optional dryRun: true for preview.

​
External TMS memories

Use your TMS or the Hyperlocalise CLI for provider-native TM download shapes where supported (csv or tmx depending on provider). CLI CSV is often Hyperlocalise-shaped and may not match every TMS upload format.

​
Tips

  • Prefer TMX when moving segments with inline tags or stable tuid values between memories.
  • Prefer CSV for spreadsheets, quick edits, or scripts.
  • Re-importing an exported TMX from Hyperlocalise is supported; treat CSV as the canonical flat row format for native memories.