Figure 42.4
Four-click sequence with body.
488
Switches
some backwards-masking effects, as the clicks are very close together and it is
hard to pick out their sequence order. A big difference in the total effect occurs
once you have three or four clicks in close time proximity.
Slide Switch
The above switch sound lacks detail. In particular, single bands of noise pro-
vide only a rough sketch of what we want, and it’s hard to find the balance
between a wide bandwidth that sounds crunchy or noisy, and having them too
tight which produces a nasty overresonant ring. We need more tailored signal
sources for our click and slide components. Let’s begin with a simple way to
approximate a short metal ping.
Figure 42.5
Shortping.
From the section on psychoacoustics you will remem-
ber that pitch and spectral discrimination decreases as
we move into high frequencies. Not surprisingly, quite
complex high-frequency vibrations like a small spring
or tiny metal plate may be approximated with only two
or three sinusoidal waves. In figure 42.5 we see a pair of
sine oscillators, which will be tuned to frequencies near
10kHz. Once again, a short square decay envelope modu-
lates them to produce a quick “ping” sound of 50ms dura-
tion. When added to a noise-based metal click this really
boosts the presence of the sound. It produces a much
more solid and focused effect. Let’s call this abstraction
and use it later.
Figure 42.6
Pnoise.
White noise sources have an unfortunate prop-
erty here. For very short sounds it’s hard to know
whether any segment taken at random will con-
tain the frequencies we want; hence there is a
random fluctuation in the level of each click pro-
duced by band-filtered white noise. What if we
were able to produce a signal-like noise but with
a more controlled spectrum? The abstraction in
figure 42.6 shows such a method. This creates a
spectrally dense signal that works like noise, but
only within a certain bandwidth. Ring-modulating
three phasors (and modulating the first stage side-
bands from each pair) gives a very “jagged” wave-
form. Frequencies are chosen to produce a dense
distribution around one point. Taking the cosine
of this signal is equivalent to summing many sine
waves centred around the base frequency so we get a similar effect to a band
of filtered noise. However, unlike a true random noise source, this “additive
cluster noise” is guaranteed to have the same strength of each partial at all
times. Let us denote this abstraction, which takes three frequency parameters
to set the noise colour, as
.
Do'stlaringiz bilan baham: |