I used a rather expensive IBM OCR for a while some time ago, that I bought to replace what came with an earlier scanner. I'll agree with RB that IBM has made some really weird sh*t, but in this case it was a little better than the other one. That was several years ago, and the current ABBYY is greatly improved over either of those, although still not fully error free. It has to be assumed that any such "transmogrifier" needs a "dictionary" of what words there are in the language you're using, in order to guess well. Maybe "upper middle class English (yUK style) isn't a mainstream language (at IBM et. al.). The Microsoft Press Book Developing International Software, in the edition for Win95 (©1995), shows 47 different keyboards (an incomplete list) and most of those can be used for multiple languages/dialects. My most spectacular recent OCR failure was on the scan of The Dictionary of Egyptian Hieroglyphics, Vol I. Even Microsoft doesn't have a keyboard for those chars, and there's no UNICODE character definitions for them, so far as I've found. UK English probably lies somewhere between heiroglyphics and other real languages. (Just takin' a poke at RB. I've heard a number of UK speakers, and some of it really was mostly intelligible.) John
|