On Wed, 16 Jul 2003 16:34:34 +0100, "Graham Holloway"
wrote:

I seem to remember reading about a Korean(?) company experimenting with a
public address system that generated a carrier around 20KHz plus an audio
signal shifted up by 20KHz, the assumption being that the non-linearity of
the ear would make it audible. The transducers were placed in a matrix
several feet above head height, and because of beaming at these frequencies
the sound(!) would appear generally from above. But the last I heard was
that they were experimenting with a membrane above the listener to create
additional non-linearity.

Or this: http://www.atcsd.com/tl_hss.html

Kal