Bat
Detectors and Bat Detection Techniques
This topic is for
anyone who has ever wondered what a bat really sounds like. The majority
of bat vocalizations are at frequencies much higher than human hearing.
We call these frequencies “ultrasonic” frequencies. In order to hear
bats, we use bat detectors. Bat detectors use various detection
techniques to shift the bat’s ultrasonic frequencies down to the human
hearing range. However, not all bat detectors and techniques are alike.
In this article, I will describe and demonstrate several of the commonly
used bat detection techniques. I will also demonstrate a more advanced
technique that illustrates the great advantage gained using Digital
Signal Processing to answer the question “What does a bat really sound
like?”
In the following sections, four bat detection techniques will be
described and compared. To be consistent, all four techniques were
applied to the same signal. I used an AR100 ultrasonic receiver to
generate a wideband recording of a group of Western Pipistrelle bats. I
then reprocessed the recording digitally using each of the bat detection
techniques to generated MP3 files covering the same 10 second segment of
the original recording.
|
Technique #1 : Frequency Division |
|
Frequency division is common technique employed in inexpensive
bat detectors. It is commonly implemented using a zero crossing
detector along with a digital counting circuit which divides
(counts) the ultrasonic frequencies down to the audio range.
It is a simple technique, but it is also destructive to the
signal. Its performance is limited because 1) it does not
preserve amplitude information, 2) it generates a high level of
artificial harmonic content that cannot be removed, and 3) it
does not distinguish well between the actual signal and the
background noise, so it is very noisy.
Click on the microphone icon for a frequency division demo, but
BE WARNED, the noise content of this technique is very high so
keep your volume down!
|
|
Technique #2 : Frequency Division with Amplitude
Retention |
|
Thankfully, the noise problem with basic frequency division bat
detectors can be improved simply by adding an amplitude
retention circuit. Amplitude retention is implemented using an
amplitude tracking circuit that operates in conjunction with the
frequency division circuit. The signal amplitude is tracked
and then reapplied to the output of the divider circuit in order
to distinguish between the desired signal and the background
noise. Its performance is better but it still does not solve
the artificial harmonic issue. This type of bat detector
generates audio that sounds more like flying castanets than
real bats!
|
|
Technique #3 : Frequency Compression (State of
the Art) |
|
Frequency Compression is an more advanced form of the frequency
division technique, but it can only be employed in bat detectors
that use digital signal processing. Frequency Compression uses
a linear frequency transformation rather than a dividing
circuit. This eliminates the artificial harmonics and creates a
more acoustically accurate result. One of the main benefits of
Frequency compression is that it maintains the harmonic
relationships of the original signal while scaling the
frequencies into the audio range. It also operates over the
entire ultrasonic range, so no tuning knobs are required!
The frequency compression demo is a preview of the real-time
frequency companding software that BAT will be releasing in Q1
2005. Enjoy!
|
|
Technique #4 : Heterodyning |
|
Heterodyning bat detectors operate like a radio receiver. They
use radio tuning techniques to translate audio sized portions of
the ultrasonic frequency band down into the audio range. The
main advantage of this technique is that it produces a high
fidelity signal. The main disadvantage is that it only
processes a small portion of the ultrasonic frequency range at a
time. The main complaint from users is that, if a heterodyning
bat detector is not tuned correctly, you will miss the bats you
are looking for.
|
|
Technique #5 :
Time-Stretching |
|
Time-Stretching is a not a technique normally applied in the bat
detectors. In general, it is more effective when applied as a
post-processing or playback process. Time-stretching is
generally performed by first recording the bats using a wide
bandwidth recorder and then playing back the recording at a
reduced rate. Think of this as “slow motion” playback. |
|
In conclusion, frequency
division is the most common type of bat detector, but it is also the
most destructive to the signal quality. It’s main benefit is that it is
simple and inexpensive, but it’s fundamental problem is that it creates
a large amount of artificial harmonics.
One step above frequency
division is heterodyning. Heterodyning produces a very high fidelity
signal, but is suffers from two problems. First, only a portion of the
ultrasonic frequency range is translated at a time. This may not be a
issue for monitoring bats like Pipistrelles, that have a narrow
vocalization range, but it becomes an issue for bats like Mexican
Free-Tail or Brown bats that have larger vocalization ranges. Second,
heterodyning does not preserve harmonic relationships, so the
translation can produce unnatural (“Star Wars”) sounding results.
The most acoustically
accurate methods are frequency compression and time-stretching. Both of
these techniques compress the full ultrasonic range into the audio
range. This scales down the frequencies while maintaining the original
harmonic relationships. Nether of these generate artificial harmonics
and both produce natural sounding results.
While I hope you found the
discussion of bat detectors and bat detection techniques informative, I
also hope that it shed some light on the real question: “What do bats
really sound like?”
|
|