Ray Dillinger wrote:
> If you get two sensitive microphones in a room, you 
> should be able to do interferometry to get the exact locations 
> on a keyboard of keystrokes from the sound of someone typing.  

Interesting.  Probably not the easiest way to snoop, but you might be
driven to it.

> I guess three would be better, but with some reasonable 
> assumptions about keys being coplanar or on a surface of known 
> curvature, two would do it.  Interesting possibilities.

Interferometry like measuring the time delay between the two
microphones?  Defines a hyperboloid, which when intersected with the
keyboard still isn't specific enough, so I think you need three mics.

> [A quick contemplation of the wavelength of the sounds in question
> would put an end to that speculation I suspect. --Perry]

You can localize to better than the shortest wavelength present, so
the spectrum isn't obviously a problem.  Consider it under ideal
conditions -- anechoic, no transmission losses, omnidirectional
emission.  Then the mics get the same signal (at different times), and
you can just find peak correlations between them.

The required accuracy is roughly a centimeter, or 30 usec of sound
travel, over one sample at audio rates; adjust that trigonometrically
for mics placed other than 60 degrees apart.  Keystrokes are noisy and
should make decent correlation codes.  Less-than-ideal conditions
might make the scheme impossible, but I don't know how to conclude
that without a lot more work.

I don't know the state of the art, but a little web searching appears
to say that people can localize speech in a videoconferencing room to
within one 44-kHz sample.  http://www.ie.ncsu.edu/kay/msf/sound.htm

     Eli Brandt  |  [EMAIL PROTECTED]  |  http://www.cs.cmu.edu/~eli/

Reply via email to