Probably not noise cancelling since it needs much of the info in the environment to retain position info. More likely a compression algorithm that instead of looking like an upside down hockey stick (with input volume on x axis and output on y), it does a combination of adjustable upward compression on lower volumes and downward compression/brickwall limiting on higher volumes. This will attenuate loud events and boost background sound. Positional data would come from stereo mic pair, one on each ear, which would approximate the way ears do locational info.
Slashdot Mirror
Probably not noise cancelling since it needs much of the info in the environment to retain position info. More likely a compression algorithm that instead of looking like an upside down hockey stick (with input volume on x axis and output on y), it does a combination of adjustable upward compression on lower volumes and downward compression/brickwall limiting on higher volumes. This will attenuate loud events and boost background sound. Positional data would come from stereo mic pair, one on each ear, which would approximate the way ears do locational info.