Hi all! i understood, the demo application has a bug. If you recognise entire document, you have possibility to check language, if you working with fields, you havn`t. This became a reason of my mistake in conclusion. Really, OCR is pretty fine.
answered 28 May '14, 08:55