I'm putting together a system for the first time, and I'm thinking about using the MSAgent SDK to write a Voice Control system for the computer. Has anyone else attempted/done this? I don't think it will be that difficult to implement a simple setup.
I have used the SDK, but not for voice recognition. Only for synthesis. Either way, I don't think that recognition will work very well in an automotive environment. And voice controlling something playing music is kind of counterproductive anyway... :)
I used to own a Clarion AutoPC and voice recognition works fine in a car. Even when playing music at normal levels it had about 98% accuracy. As soon as my system starts coming together I'll be using VR
But the Speech SDK is not designed for an automtove environment. Perhaps training it on the road would help? Hmm...
I am using a program called Game Commander 2. This program is very limited in that it only works with your keyboard. You tell it what to listen for, when you say that word(s) the program presses the key(s) you programed it to press. However, it does recognize almost any voice or word with little or no training and only costs about $30.00 US.
It works very well for me and I get the added plus of an easy to program voice recognition program for my games. In older versions they had a sleep/wake-up feature. This allowed for you to tell it when to "listen" and when not to. They are adding this feature back into the new version with a patch which should be out within the month.
I have it set up to do everything that can be done with winamp.
It also has "global commands" so that you could close out some software (winamp) and start other software (dvd) then use a different command set to control the dvd. Sounds harder than it really is.
Website is http://www.gamecommander.com/products/gc2.html