deloptes <delop...@gmail.com> writes: > Martin McCormick wrote: > >> I have 4 older PC's that generally work well running >> debian but Right now, 3 of them need varying degrees of attention >> to their BIOS setups as Dell motherboards and possibly other >> brands will occasionally modify their boot sequences for some >> reason and the only way one can boot from a CDROM is to get in to >> the BIOS setup and yank the boot order back to one where the CD >> drive is ahead of the hard drive or put an unbootable hard drive >> in. Six or eight months later, one will suddenly discover that >> the boot sequence has fallen back to the useless one where the >> floppy drive is first, followed by the hard drive followed by the >> CDROM. > > I have not heard so far of working OCR free under linux. I would buy > commercial software that can work as screen reader. You can pipe the images > from the linux PC to that software and hear the text. > > This is not only the OCR part, but also the reading and what I have seen > under Linux is all BS. There were very good projects 10-15y ago, but I have > not heard of a break through in any of them. For example ViaVoice a STT > program used to run on linux and was dropped when IBM and Phillips stopped > the project cause DARPA stopped the funding. Festival is good as TTS, but > is far away from commercial quality and so on. > > Making a good reader program is very complex and sophisticated thing and > there are many unsolved problems to deal with. This means you can not just > OCR the screen and dump it to the reader - you must preprocess it and often > go to semantics, which makes it pretty difficult.
I used tesseract-ocr, mentioned previously, a couple of years ago with very good success. Also, the problem he's trying to solve is much simpler than the general OCR problem; he's got the actual correct pixels (rather than a scan), and maybe even have knowledge of what fonts are used. That makes a huge difference. > If you have time to waste, keep us posted of your progress. Last time I did > research on the topic was 10+y ago, when I wrote my thesis on dialogue > systems, so please, correct me if I am wrong and if there is a useful > software out there. Honestly I doubt it, cause people massively got dumber > in the past 10y, of course except the readers of this list :) > > best regards
