Biomedical Engineering Reference
In-Depth Information
The cyclic gestures giving rise to syllables can be generated with a second
oscillator with the same design as the one used to emulate the labial motion,
but with
R
1
≈
2 MΩ. This allows the device to run on 9 V batteries, and
avoids the use of a function generator.
7.3 Playback Experiments
One of the purposes of synthesizing song is to build confidence in our mod-
els. If, by prescribing mathematical rules for the dynamics of the avian vo-
cal organ, we can emulate a behavior leading to realistic sounds, we build
confidence in our having identified the appropriate physical mechanisms un-
derlying vocal production in birds. Beyond this satisfaction, there are other
interesting possibilities for these techniques.
A complete topic could be written on the communication between birds.
We are focusing in this topic on the processes that they use to vocalize, but
what do they “say”? There is a wide consensus on birdsong being important
in at least two different behaviors: territorial defense and mating. Field scien-
tists test their ideas concerning these issues with a variety of techniques, one
of them being playback experiments. For example, a recorded song may be
used to persuade a bird that an intruder is in its territory. The use of either
numerical or analog simulations instead of recordings widens the range of
acoustic features that could be used in playback experiments, and provides
the experimenter with the possibility of experimenting with sounds well be-
yond what can be achieved by simply editing recordings.
Simulations can definitely play a role in experiments on learning and au-
ditory feedback. A computer connected to a loudspeaker can easily play back
a huge number of qualitatively different songs according to a prespecified
rule, throughout the learning period of a newborn songbird. There is no need
for previous song recording, nor subsequent classification and storage. Fur-
thermore, with an electronic syrinx capable of analog simulation, the experi-
menter now has the possibility of changing detailed features of the simulated
song in
real time
. Auditory-feedback experiments involving altered syllable
duration, altered frequency range and altered frequency evolution (even with
a downsweep changed into an upsweep) are available.
7.4 Why Numerical Work?
Most of the theoretical results discussed in the previous chapters were ob-
tained by numerical simulations of the equations describing the physical
mechanisms involved. In the previous section, we showed how to proceed
by generating of synthetic data. Is it always necesary to rely on numerical
tools to advance our understanding of the generation of birdsong?
