I’m curious to know how this model tackles newly released fonts. How difficult would it be if this model needs to recognise font in a different language?