If you’ve tried to use speech recognition to control your computer, then you may have experienced the frustration of finding “George, stop touching that!” transcribed into your letter to the bank since the system doesn’t differentiate between your dictation and your asides. Intel are hoping to put a stop to all that, with their integration of facial-recognition and speech-recognition; part of the company’s Research team experimentation, the prototype uses a webcam to track facial movement and only turn on speech-control when the user is facing their monitor.
Video demo after the cut
That way, you can maintain a conversation with someone in the room without needing to shut off the microphone, then turn back to the PC and ask it to run a Google search or transcribe a letter. Intel also envisage the face tracking system to be useful in gaming and video conferencing, as it can accurately (and with low overheads on multi-threaded systems) identify the user’s face and crop down the picture to just display that or even use expressions to control gameplay.
The system uses the native speech recognition that’s included in Vista and will be included in Windows 7; however there’s no talk of releasing the video tracking portion. More details in the video below.