Zeddy, a Microsoft Academic Developer Evangelist in Indonesia, has a nice demo of using Windows speech recognition to remotely control a robot. The application is running on a laptop which controls the bot wirelessly. The voice recognition happens on the PC and the instructions are translated into commands for the robot to execute. The math commands are executed entirely on the laptop.
Video: Zeddy demo's Windows Vista speech recognition to control robot
It is still very hard to achieve remote voice control were an open microphone or array of microphones can pickup voice commands given in the normal domestic or work place. Windows Vista has support for microphone arrays built in, but the focus of this support is for the type of microphone array to be built into laptops.
One of the tremendous benefits of working at Microsoft for a hobbyist like me (see my robotics blog here) , is that I get to discuss my project dreams with people that really know the reality. I was able to ask Ivan Tashev - software architect at MS - about his development of the Windows Vista microphone array technology. Ivan says 'Windows Vista supported microphone arrays work in the zone of +/-50 degrees, while for the robot will be better to have 360 degrees listening zone – something like this.' It was this Microsoft Research (MSR) picture that got me thinking I could do remote voice recognition with A1.
So onto my plan B which is to use a form of wireless headset such as a bluetooth or DECT directly against A1, but again success will depend completely on the quality of the voice transmission. If I can't make this work satisfactorily, plan C is to do the same as Zeddy. Using MSRS you can easily have distributed services aggregated to a single application. So A1 could have all it's drive and sensor services locally, but make use of a remote voice recognition service run from a laptop or PC. My plan C would be to have a UMPC running my voice recognition MSRS service which is aggregated into A1 via the wireless network. This would enable all the voice commands to be processed on the UMPC and just the code commands pushed to A1.
Still, I have navigation to implement before I get here, but I like Zeddy's demo - cool stuff!! Remember voice recognition and speech synthesis are all there in Windows Vista waiting for you