Owner`s manual

16
CHAPTER 3 -
The Craft of audio Synthesis
CHAPTER 3 -
3.3
Attributes of Auditory Events
Of all the thousands of sounds you hear in a single day, only some
have a de nite beginning or ending point. Many just come and go
without a clear start/stop. Those that do have a reasonably clear
start/stop we will call auditory events. The other sounds, the less
clearly de ned ones, we can call auditory
textures
or
backgrounds
or
even
structures.
Think how some sounds are more complicated than others. The
hum of a refrigerator is sort of simple, the hum of a tuning fork is
(as we heard above) really simple. A single bass note from a piano
is quite complicated, the way it keeps changing and evolving as
it dies away. Human speech is a very complicated kind of sound,
even when it’s your native tongue (when it’s a language you don’t
understand, it sounds even more complicated).
There are many, many more kinds of sounds to hear than you or
I have words to describe them with. Here are just a couple of the
most general distinctions people make about different kinds of
sounds:
3.3.1
Steady-State Attributes:
some sounds, once they get started, remain pretty constant.
They don’t change much while they continue, and when they stop, they just stop. We say
such sounds reach a
steady
state, in which we can pick out a de nite
steady state, in which we can pick out a de nite steady
Pitch, Loudness,
and
Tone Color.
Notes from organ pipes – or from electronic drawbar organs – are steady-state
sounds.
3.3.2
Time-varying Attributes (i.e. “Envelopes”):
some sounds have a pretty clear start
and end, but while they’re happening they change. Think of a bird song, or ordinary
human speech.
Such sounds can sometimes be analyzed as more or less rapidly changing in one or
more of those fundamental auditory attributes: pitch, loudness, or tone-color. Think, for
example, of how the sound of a plucked guitar string evolves from the moment you pick
it to the moment you can’t hear it anymore. It starts out loud and fades away; and it
starts out “bright” - with lots of harmonics – and is slowly “muf ed” as it fades out.
One way to conveniently diagram what happens in such events is to chart each changing
attribute separately, in its own time-domain graph.
Such a graph is often referred to as an
envelope.
Over the years, some standard vocabulary
has developed for talking about envelopes:
Attack
(for the  rst part of an event),
Decay
(for
Decay (for Decay
some later parts), and
Release
(for the last part, when the event is coming to an end).
“white” noise has equal energy in any
two equal frequency bands…
0dB
-30dB
so it’s NOT equal in equal pitch intervals…
…even when you use dB for the vertical axis…
THIS is why white noise sounds so shrill
octave intervals
20
4
-12
-9dB
-6
-3
6 8
10 12 14 16 18 20K
20
40 80 160 320 640 1.25 2.5 5 10K 20K