Subject: Re: palm top pipe dream
Date: Mar 24 09:40:17 1995
From: Alvaro Patricio Jaramillo - jaramill at sfu.ca


>
> For routine bird censuses, for years I have suggested putting up a series
> of microphones/recorders in all sorts of habitats year after year to do
> bird censuses. You'd have to hire people to enter the song types and
> frequencies in a database (employ more biologists/naturalists!), but it
> would be most efficient to have song-recognition software that would record
> species automatically, beaming the signals to a central computer. What a
> way or an ongoing assessment of bird populations. But nobody ever listens
> to me.....sigh.

This is already happening! In Ontario, a system is being tested that
remotely records sounds out in the field and feeds them to a computer.
Specially written software uses neural networks to compare the signal to
templates of typical calls stored in the computer. This system has had
a 98% accuracy rate with clean signals of frog calls. I read a note about
this system which outlined some of the features:
-transforms field data into digital format
-data can be presented as both time history or in frequency domain (spectro
gram).
-ability to extract pertinent statistical information about the signal
such as power, mean amplitude etc.
-ability to reduce data sets to a more manageable size without much infor-
mation loss (such as dominant frequency, average strength).
-providesw a verification so that any digitized audio signal (whole or
part) can be expressed as an audio output and/or viewed on a computer
monitor.

Apparently they have also designed a software package that can calculate
relative abundances of different species by comparing the pulse width,
and voltage characteristics of the vocalizations.

So I guess someone is listening to you! :->

Al Jaramillo
jaramill at sfu.ca