Tuesday, February 21, 2006
Tamarih is off to Mali again, though this time it is for five weeks. This gives me plenty of time to study for my two chinese courses at UQAM, as well as play around with some geeky things. The main one bugging my brain right now is 'speech recognition'.
Last year, I had the chance to try out a Toshiba Tablet (M40 I think, the high-res 12" one) which was running XP tablet. It featured some speech recognition controls that worked really, really well. So, I felt it was time to get something running that I could program with. Nothing to free-form, just some silly things like getting a computer in the house to control lights, or change music, or answer the phone. Eventually I'd like something to be able to answer questions like, "What's the weather supposed to be like today?" or "When is the next bus on the 144 line?".
Anyway, I'm starting with the CMU Sphinx code, but this is not trivial to set up. It seems that despite it being open source and all, that no one has documented any successful 'hacker' implementation of a running system. Perhaps I'm doomed.
I'm starting with the tutorial, but I think I would like to skip the training and try to make use of the opensource acoustic, language and other models. If I knew what the hell they did.
Last year, I had the chance to try out a Toshiba Tablet (M40 I think, the high-res 12" one) which was running XP tablet. It featured some speech recognition controls that worked really, really well. So, I felt it was time to get something running that I could program with. Nothing to free-form, just some silly things like getting a computer in the house to control lights, or change music, or answer the phone. Eventually I'd like something to be able to answer questions like, "What's the weather supposed to be like today?" or "When is the next bus on the 144 line?".
Anyway, I'm starting with the CMU Sphinx code, but this is not trivial to set up. It seems that despite it being open source and all, that no one has documented any successful 'hacker' implementation of a running system. Perhaps I'm doomed.
I'm starting with the tutorial, but I think I would like to skip the training and try to make use of the opensource acoustic, language and other models. If I knew what the hell they did.
