Comment by intelkishan

I guess the lack of training corpus would be a major issue with such a use case.