On Thu, Oct 28, 2010 at 5:13 AM, ss <cybers...@gmail.com> wrote: > my requirement right now is for some > form of technological device that will give me instant OCR and translations of > Sanskrit and Kannada texts
Technically, this is two problems: OCR and translation. The latter is easier to solve than the former. 1. OCR-ing of Indic texts is a hard problem. In my guesstimation, OCRing of printed Sanskrit will probably be available in the next 5-10 years (considering how Sanskrit and Hindi have almost identical scripts). Kannada is going to take longer. Just forget hand-written texts. 2. Translation is easier to solve. You need a large parallel corpus (i.e. the same text in two languages). For large languages like Hindi, French, etc., it is easy to find such large corpora (documents from the UN, EU, Indian Parliament, etc.) in a digital format. For languages like Sanskrit, such corpora exist, but are widely dispersed and in analog format. I have no idea about parallel corpora for Kannada. My bet for something like your dream device is a smartphone. If you have access to an Android phone, check out the Google Goggles application. Here are a couple of demos of the app: http://vimeo.com/15024221 http://www.youtube.com/watch?v=GgcE_EQRpdA It is not exactly what you have in mind, but it certainly points in the right direction. Thaths -- Marge: Quick, somebody perform CPR! Homer: Umm (singing) I see a bad moon rising. Marge: That's CCR! Homer: Looks like we're in for nasty weather. Sudhakar Chandra Slacker Without Borders
