Things are not looking good for Microsoft’s Vista, especially when it comes to listening to music and surfing the web at the same time. Slashdot Link.
It uses the microphone to detect echo from your head. This starts with the first approximation that your head is symmetrical, smooth, and round. If the echo shows any sign of left/rigth asymmetry, it brings in the next layer of feedback control by simulating a rotated ovoid head, and progressively brings in more features such as topological variations (nose, eyes, ears, open mouth). It is continually trying various time delays to make sure it isn’t confused by emenations from your own mouth, nose, or ears (tintinabulation).
Once it determines the maximum quality feedback parameters, it backs off various parameters to try to reduce the computational footprint. It keeps a record of these adjustments and periodically adds them back in temporarily to make sure the basic parameters are still valid. If any of these trials show the need, it will restart the complete feedback search cycle.
Where does the network figure in all this, you ask? Simple. All that I have described so far is reactive feedback. An ounce of prevention is worth a pound of cure, or more usefully, predicting how much feedback control is necessary can pay bigger dividends — more bang for the buck, so to speak — than reactive analysis. If it can tell what you are doing from packet analysis, it has a better chance of predicting your head position. It looks at HTML pages and tries to guess what content is shown, in order to know if it is likely to affect your head position, and then tries to guess where that content will show on the screen, in order to predict where your head will be.
Coupled with mouse and keyboard controls, this can lead to amazing sound quality from the piss-poor speakers found on most laptops, even simulating 5.1 speaker systems with just the two speakers found on most computers.
Now you know.