Out of curiosity: Is the first OCR example really the best you can find? Is there no open source solution that outputs good results for handwritten notes?
There are open source "computer vision" libraries which work really well, but, also there is a file on your filesystem with all the words, so you could pass over the OCR'd text to fix "typos".
