Some Audio Puzzlers
In the letters I get from readers of various publications,
and this site, certain basic questions arise time and again. Here are a couple of them.
First is the confusion about decibels, and how a level
difference of 3dB represent a doubling or halving of power in some cases, while in others
it means just a slightly noticeable difference in volume.
Tricky things, decibels. One thing a decibel is not is a
unit, like a watt or a volt; it expresses a ratio between the intensity of one signal and
another.
In the electrical part of an audio system, such a ratio
could be expressed many ways. When comparing a 25-watt amplifier with a 50-watt amplifier,
say, we could say the ratio was "2:1" or "twice" or "2x" --
any would serve equally well. But the result of everything that happens in an audio system
has to be meaningful in terms of the final element: the listener. Unfortunately, human
beings do not perceive level differences in a linear manner -- doubling the electric
signal to a speaker will not result in an apparent doubling of sound level. In fact, the
ratio of the softest sounds we can hear to the loudest we can endure is about a trillion
to one.
Our perception of level changes is logarithmic, so a
logarithmic notation has been adopted for use in all parts of audio. This has the
advantage of keeping the numbers relatively small (that trillion-to-one translates as
140dB), and allowing simple computation of signal levels. To multiply two power
differences, all you have to do is add or subtract the ratios in decibels.
All of this can be very confusing to us ordinary mortals
who are unfamiliar with this sort of math. But it simply means that, in order to
accomplish relatively minor apparent sonic changes, some fairly major electrical ones have
to take place.
Much of the original establishment of the decibel scale was
very subjective because it is extremely difficult for anyone to know for sure that one
sound is exactly, say, twice as loud as another. But after a great deal of
experimentation, averaging the responses of many listeners, one decibel was set at the
minimum perceptible change in level, with no reference to what had to happen electrically
to produce that change.
Building on that, it was determined that 3dB -- still only
barely hearable -- represented either a doubling or halving of electrical power, depending
on whether the change was an increase or decrease, and that an apparent doubling or
halving of level required 8 or 10dB change (10dB is the figure usually quoted because,
electrically, it neatly represents a ten-fold change).
None of this would be very important to anyone but
engineers were it not for the fact that it bears directly on the sort of amplifier power
you should have for true high-fidelity sound. To give an example, if your system coasts
along at about 2 watts for average program level -- by no means an unusual situation --
and then hits a very modest 20dB musical peak, all of a sudden your amplifier will be
called upon to deliver 100 times the power (ten plus ten decibels means ten times
ten watts), or 200 watts. A really heavy transient will require a lot more power than
that.
Another puzzler is how the tiny diaphragms of a pair of
miniature headphones or earbuds can produce the whole musical spectrum, from organ
pedal notes to tinkling triangles.
Your eardrums can respond to such a range, so why is it
surprising that a device with perhaps ten times the surface area can produce it?
Conventional audio (that is, using speakers) must contend
with a lot of physical laws that conspire to degrade sound before it ever gets to your
ears. For instance, in a real acoustic environment -- a room -- high frequencies do a
couple of things. First, they "beam" -- they are directional -- so to the extent
that you are even slightly off a tweeter's axis, treble will be attenuated. Second,
high-frequency signals naturally "roll off" the farther away you get from the
source, so unless you are willing to sit with your ear pressed against the speaker, you
will perceive less treble than the speaker is producing, compared to the rest of the
spectrum.
With headphones, you can't be off axis; the phones are
pumping the signal directly into your ears. Neither do you suffer from the roll-off that
comes with distance; the transducer is typically less than an inch from your eardrum. In
terms of high-frequency energy, headphones receive exactly the same signal as speakers,
but you get all of it.
At low frequencies, speakers produce an enormous amount of
wasted energy. Bass is essentially omnidirectional, radiating equally in all directions.
The amount that finally reaches your ear is a minuscule percentage of the total, which
means that speakers must put out a prodigious amount of low-frequency energy for you to
hear adequate bass. The tiny element in an earphone produces very little energy at the low
end, but since the phone is closely coupled to your ear canal, very little is lost to the
surrounding air.
With speakers, the acoustic environment has an effect on
sounds at all frequencies. Furnishings can absorb or reflect treble (typically, they do
both at the same time, selectively}, while bass is subject to influence by the physical
dimensions of the listening room (standing waves can boost or cancel certain frequencies)
as well as the limitations of the speakers themselves. Headphones have no such restraints
-- what the diaphragm produces, you are likely to hear.
Even so, headphones are far from perfect reproducers of
sound. Certainly they are able to produce very strong treble and surprisingly robust bass,
but that doesn't mean that the frequency spectrum is balanced, however spectacular it may
seem.
An earphone sets up a pressure system in the outer ear that
is quite effective at transmitting to the eardrum exactly what the transducer is producing
-- if it's capable of technically flat response, that's what you will hear. Unfortunately,
the ear itself is not a flat device. All those twists and turns in the outer ear have a
dramatic effect on what we hear; a flat signal reaching the outer ear is profoundly
modified before it reaches the eardrum. It sounds flat to us because that's what we're
used to -- all sounds are modified the same way -- but on a response curve, the signal
reaching the eardrum would look like a map of the Yucatan Peninsula.
Headphones bypass all of that, so to sound natural their
response curve would have to approximate very closely the net acoustic effect the outer
ear has on a flat signal. This can be approximated but never duplicated because no two
ears are alike.
To complicate things further, most recordings are
engineered to be played through speakers in a real acoustic environment. Headphones give
such material an exaggerated, wider-than-life sound, characterized by the
"soloist-in-the-middle-of-the-head" effect. This has its appeal but it's not
accurate.
Headphones can certainly do remarkable things, and they
have a very important place in audio. But they should be appreciated for what they are:
convenient, pleasant, portable, but -- so far -- specialized devices with definite
limitations.
...Ian G. Masters
ian@mastersonaudio.com
|
