I used finereader a lot for school work. It's a great program if you need to have text from pictures/scans. But it does make a lot of mistakes. Maybe with English language it works better, less complicated characters.
I recently scanned an entire English novel (300+ pages) in PDF with FineReader 11. I was completely blown away with how few
mistakes it made recognizing characters/words. Really quite exceptional.
I'm sure it's probably a lot different with languages that use a lot of diacritics, but I'd think even that
would be something that could be overcome by creating/training your own recognition dictionary. The ability to easily train
Finereader's recognition routines are invaluable, but probably largely under-utilized.