> why should a preamp be called on to amplify frequencies above 20 K if nobody
>can hear it?
That is an excellent question!
First we must distinguish between recordings made for analysis and recordings
made to be listened to by people. If your purpose is to assess the frequency
range of Bat vocalizations then you had better have equipment that can record
it. But you are not talking about that case.
There are several things to remember. The first is that the frequency range of
human hearing varies quite a bit between individuals. As a rough
approximation,
people with smaller ear geometry have a frequency range that is shifted up in
frequency relative to the average. The second is that the threshold of
hearing,
the level of the quietest sounds that can be heard, is not at all flat.
Finally, there is the important effect of auditory masking, which is probably
most significant of all.
It's worth taking a look at a typical threshold graph:
http://en.wikipedia.org/wiki/Equal-loudness_contour
This data is from an ISO standard, ISO 226:2003. It's a bit of a moving target
because different labs in different places get different results, and in any
case this curve is meant to represent an average result. But an important
thing
to note is that the thresholds get worse (higher) at low frequencies and at
high
frequencies. Compared to earlier editions of the standard the present one
doesn't have data for frequencies above 12.5 kHz. This is because there is too
much variation for them to come up with a representative figure. The dotted
lines are an extrapolation.
You can see this a bit better in Figure 1 of this paper by the folks at
Etymotic Research:
http://www.etymoticresearch.com/publications/erl-0096-1997.pdf
You will see that they show the threshold going pretty much straight up as the
frequency approaches 20 kHz. Hearing pretty much stops working at some very
high frequency. The frequency at which this occurs does vary from individual
to individual. In "Recommended High-Frequency Audiometric Threshold
Levels (8000-18 000 Hz) ", JASA Vol52, No. 2, The authors obtained a mean
threshold value of 59 dB at 18 kHz. They didn't obtain data for higher
frequencies.
At low frequencies the behavior is completely different. The threshold keeps
rising as the frequency decreases but there is no clear cutoff. So basically
the 20 Hz to 20 kHz frequency range is a misnomer. If one were to define the
frequency range of human hearing as the range which we can hear at a level of
60
dB SPL then, using the data from the above sources, that range would be about
30
Hz to 18 kHz.
Finally, there is the effect of masking. If the sound being recorded is
complex
and has lots of overtones then the louder, lower frequency sounds "mask" our
ability to hear the higher frequency sounds. This is why we can't hear the
phone ringing while we're in the shower; the noise of the shower produces a lot
of masking.
Backing away from all of this analysis for a bit, it's really easy to make
preamplifiers that amplify 20 kHz, or 40 kHz, or 200 kHz. But it's very
difficult to make microphones that do that. And even if a microphone is flat
up
to 20 kHz on the axis of the microphone, it almost certainly has a very narrow
polar pattern at those high frequencies. So it's quite conceivable that a
microphone that is "flat" to 20 kHz actually transmits very little sound from a
subject that is off axis, effectively making a 10 kHz or 5 kHz microphone.
Eric Benjamin
|