Accurate OCR for mixed-language documents
Many business documents are not monolingual. Invoice headers in English sit beside Persian body text; Arabic contracts include Latin identifiers. Treating every page as a single script leaves gaps.
Textora exposes explicit language pairs — eng+fas and eng+ara — so the OCR engine can score both scripts instead of forcing a false choice.
On the product side, the workspace respects document direction when you switch the UI to Persian or Arabic, while keeping developer strings and curl examples stable in English for API consumers.
If you are wiring this into a pipeline, submit the same language codes via multipart language on POST /api/ocr/jobs, then poll until the job reaches a terminal status.