Jazz2K said:
Ohh I hope you understand the amount of work necessary to do what you're asking. It's not just speach recognition but sentences and meaning of groups of words put together. There are so many way to say something that it would take a lot of ressources and time to do that. I hope you don't expect game companies to develop something like that and still sell games at 60$ you're asking too much. What I think is that companies like Microsoft have all the interrest to develop this kind of feature. It could be sold to other businesses that have nothing to do with videogaming. Simulating a conversation with a keyboard isn't the same as actual people talking. There are so many accents, even us humans have a hard time understanding different accents imagine a program. Implementing voice recognition into a device like Kinect is a first step... look how much time it took to get it to other languages, speach will take a long time to achieve. When more and more devs use it in games it'll be more relevant to develop it. Unless Kinect has a speach learning program that fully understands sentences don't expect this to come to gaming anytime soon or you can expect poor speach recognition where the program will understand a few words and will extrapolate what you tried to say. In the end, devs do what they can with what they got, when they'll have 3-4 years of experience with Kinect then we'll be able to see what the device really can do. Until now, I'm glad I have more options. If we were only goint to praise and expect full maturity in everything we wouldn't be where we're at because people don't want things to advance... pew pew pew... |
I know the amount of work involved, I studied natural language parsing in university in '96 as part of the artificial intelligence course. There we made a parser to be able to make queries to an sql database using natural language. It's not that hard, the main problem is it's very language dependent as grammar varies wildly between different languages.
You don't need a very sophisticated engine though, recognizing keywords is mostly enough for games. You don't need to start with an engine that can recognize sarcasm, looking for key words like why / where / how + object is a great starting point. If you can use Bing with kinect, then it should also be good enough for game input. NPCs could always say, sorry could you repeat that?
Maybe I'm overestimating how good the voice recognition is in Kinect? Anyway replacing button presses with voice commands will make many people shrug their shoulders and say nothing new to see here.







