We are using the current version of IEVision and we can OCR with many languages except for Arabic. When this method is executed with engine type Fast or Accurate an exception is raised.
result := IEVisionLib.CreateOCR(pAnsiChar(AnsiString(_OcrLangPath)), pAnsiChar(AnsiString(_OcrLanText)), IEVisionOCREngine);
you can use any file from the OCR demo, just include the ara.cube.* and ara.traineddata files and select Fast or Accurate. I send these files in a separate email.