--- In Gianni Pavan wrote:
> great accuracy is for example required to measure delays
> among a same sound received by multiple receivers.
How far apart? You can always just string cables from a common clock
source (increasingly tricky as the runs of coax increase in length).
As Rich Peet notes, you can record (at each station) a clock, but this
may be a high price to pay (2x the bandwidth).
It may be simpler in the end to just build a "clicker". Place it at a
known location, and it makes a known, brief, noise every now and then
(ideally, the noise should be out of the bandwidth of your sounds of
interest). Since everyone hears it, you know where the receivers are
and the position of the clicker, you can caliberate out the delays,
phase errors and other bizarre effects not even known, and use the
resulting model to solve for the position of any unknowns heard.
You'll probably have to run calibrations on an ongoing basis because
of environmental changes anyways, even with a fancy ass network of
GPS-locked oscillators and such.