Translation and localization companies
Short answer: Translation companies often need to put translated text back into the original positions on client scans while promising that client files stay private. EnjoyPDF does the back-filling and checking on the translator's own computer and attaches an operation record to every delivered file.
Three common problems for translation teams
- Layout: Clients want the translated version to match the original page by page. After reflowing, tables and line spacing no longer line up, and rework piles up.
- Confidentiality: Client contracts usually forbid uploading files to third-party online tools.
- Proof of delivery: After delivering a large batch of files, it is hard to say exactly what was changed and whether each change is correct.
How EnjoyPDF handles it
- In-place back-filling: Only the characters being translated change; all other pixels keep the original image.
- Local processing: Recognition, replacement and saving of scans all happen on the translator’s computer.
- Read-back and records: Each back-filled item is recognized again in the output file and checked. Each file gets an operation record (manifest.json) with the SHA-256 of the input and output files, the number of rules, and the verification results.
Recommended workflow
- The project manager confirms font, alignment and overflow settings on one or two sample pages.
- The translator generates or imports a “source → translation” keyword table, checks it, and runs batch replacement.
- The reviewer uses the read-back results and checks the “Needs review” items first.
- Deliver the output files and the processing report.
Language coverage
Chinese, English, Thai and Japanese are built in. Vietnamese, Korean, Arabic, Devanagari and others are available as downloadable language packs. The current version does not yet support Arabic, Hebrew and other right-to-left scripts, either in the text to find or in the new text. Reliability on scans differs by script; see the language support matrix.
FAQ
Can we use our existing termbase?
Yes. For AI translation you can supply a "source => target" glossary, and its entries are used as given. Without AI, turn a bilingual table exported from your CAT tool into two columns and import it directly into batch replacement.
Our client requires that files never leave our network. Is that possible?
Yes. Document processing does not go online. Cloud AI translation sends the text to be translated to the provider you chose; if your client does not allow that, choose a local model, or do not use AI at all.