[NatureNS] Automated Twitching

Date: Tue, 21 Oct 2008 08:01:46 -0300
From: Don MacNeill <donmacneill@eastlink.ca>
To: NaturenNS <naturens@chebucto.ns.ca>
Precedence: bulk
Return-Path: <naturens-mml-owner@chebucto.ns.ca>
Original-Recipient: rfc822;"| (cd /csuite/info/Environment/FNSN/MList; /csuite/lib/arch2html)"

next message in archive
next message in thread
previous message in archive
Index of Subjects


We are being replaced!

Don

Don MacNeill
donmacneill@eastlink.ca




- AN ARTICLE FOR YOU, FROM ECONOMIST.COM -

AUTOMATED TWITCHING
Aug 27th 2008


A new device may take the sitting around out of bird-watching

ORNITHOLOGY is one of the few branches of science to which amateurs
still make an important contribution. One reason is that there are lots
of bird-watchers around to collect the screeds of distribution and
habitat data that the science of ecology relies on. And one reason that
there are so many bird-watchers (as opposed to, say, mammal-watchers)
is that birds tend to advertise their presence in a way that most
animals do not. Many have showy plumage, and many (not always the same
ones) have mellifluous songs--meaning that a lot of bird-watching is
actually bird-listening.

Neither bird-watchers nor bird-listeners can be around all the time,
however, so to make the process more systematic Daniel Wolff of the
University of Bonn is trying to build a bird-song-recognition system
that can sit in a piece of habitat and listen for the calls of
particular species until its power runs out. And it seems to work. He
recently conducted a trial of the system in an area of lakes to the
north-east of Berlin, where it looked for Savi's warbler, a small and
rare bird that loves reed beds and sounds rather like a cricket, and
also for a more common species, the chaffinch.

Dr Wolff programmed his system by plundering the Archive of Animal
Sounds at Humboldt University in Berlin--where the calls, growls and
chirrups of 110,000 species are kept. Having obtained the songs of his
chosen species he applied a mathematical procedure called a Fourier
transform to them. This decomposes a complex waveform into its
underlying components--and a sound, however complex, is simply a wave
in the air. The brain, avian or human, does something similar
naturally. The result, whether performed by transistors or nerve cells,
is a pattern that certainly differs from species to species and may
differ from individual to individual.

Dr Wolff's bio-acoustic monitor is thus quite capable of distinguishing
warblers and chaffinches among the din of insects, frogs, rain and wind
that constitute the auditory background of a marsh. The next task, he
says, is to make the system versatile enough to program itself by
recognising what is and is not birdsong, regardless of the species. If
that were possible, the device could catalogue all the distinct songs
it came across, chalk up the species that it recognised and signal the
arrival of a bird that was not in the database if no match could be
found. In that case, a search of the Humboldt archive would be in
order.

Which is all very good and scientific, and will undoubtedly aid an
understanding of natural habitats and how they change. But romantics
may be pleased to note that the human ear is still superior in some
ways. Dr Wolff doubts that even his super-dooper self-programming
system would be able to recognise the song of the nightingale. It is,
he says, too rich. Fourier, apparently, cannot transform everything.





See this article with graphics and related items at 
http://www.economist.com/science/tm/displaystory.cfm?story_id=11999753



next message in archive
next message in thread
previous message in archive
Index of Subjects