On Sat, 29 Jan 2000 [log in to unmask] wrote:

> Cheap software is useless for this purpose.

If I understand the object of this analysis correctly, there is
really no need for complex software.

What we are looking for are certain diagnostic frequencies in
the spectrum of bee sounds. The tough stuff should be only the
initial identification of the frequencies that indicate certain
kinds of activity. That apparently has been done, but can be
repeated with instruments as basic as oscilloscopes. From that
point, all that is needed is a filter that will become active
only in the presence of the identified frequency -- or at most,
a set of filters responding to the several distinctive frequencies.

It was about 25 years ago that I was involved with a project intended
to recognize certain features of human speech. We used a "desktop"
computer which occupied a 6-foot long table. It read perforated paper
tape, and had all of 4K of memory on board. There was an array of
sound filters activated by distinct ranges of frequency. The software
consisted of simple "if yes, then x; if no, then y" statements.

It was no big deal.

Susan

--
Susan Nielsen                   | Beehive: If you build it,
[log in to unmask]            | they will comb.