Behaviour, Structure and Causality in Procedural Audio - Game Sound Technology and Player Interaction: Concepts and Developments - page 324

have been, and are still being, developed to spice
up sampled sources and impart random features
that seem to imitate a behavioural source. Unfor-
tunately for this approach, the underlying features
indicative of identifiable behaviour are far from
random. Bregman (1992, pp. 10-36), and again
Vicario (2001), point out that our sense of realism,
in terms of confidence in a concrete identification
with well-formed behaviour, increases with the
number of instances whose variance reveals an
underlying model (the more examples of a face
we see the better we get at identifying its true
essence). So, contrary to the goals of “increas-
ing realism through variation”, resynthesising
or treating samples to obtain random dispersions
of attack time, phase, and pitch may just lead to
a muddy confusion at the cognitive level, like
seeing several distorted and falsely coloured pho-
tographs of the same face in the style of Warhol.
No underlying behaviour that can reveal the thing
in itself is conveyed.
So, to answer the question “how real is it?”,
I postulate two different kinds of realism. There
is a superficial kind of realism, that works from
one angle and relies on smoke and mirrors, and
there is realism in depth . I also call the former
type of surface reality aesthetic or sensible realism
and the latter type of realism behavioural, char-
acteristic, or essential realism . Sensible sources
just provide the right sensations, while essential
sources provide the correct perceptions, in the
examined case. This is something obtained for
free by a source (as opposed to signal) method.
A twist to behavioural approaches, as we shall
see, is that you can fake the parametric depth to
a degree you can comfortably afford, in constant
memory and CPU space.
According to Warren (1992), Bregman (1992),
and those of the cognitive schools, behavioural
parameters are as important in our assessment of
reality as surface features. We are accustomed to
having our senses trick us with a small number of
sample points, thus perception is a multi-layered,
iterative, convergent affair: a cognitive integra-

tion. Whether multi-modal or through repeated
examples through the same channel, perception
in depth will only stand up if there is behaviour
in depth. This can happen on a short timescale,
on the acoustic and neurological time-scales in
which the sounds of a snapping twig (potential
predator) and a raindrop can move from sensa-
tion to perception, and then to identification in
the conscious awareness of the fore-brain. Or it
can happen over a medium timescale given sev-
eral instances of a source held in echoic memory
when we are capable of discerning fine nuances of
quality, formant, and scaling (qualitative feature
recognition, see McAdams & Bigand, 1992).
To discern a Steinway piano from a honky-tonk
usually takes more than one note. On still longer
timescales that involve stronger memory forma-
tion, a collection of asynchronous sample loops,
say of windy weather, may fool us for many
minutes or hours before we become aware that
the behaviour lacks depth.
Such discernment may even require the gamer
to engage in many sessions of play before the dif-
ference between repetitive, data-driven sound and
“living” procedural sound is apparent. Here lies
a vital development metric, if you never expect
the player to play the game more than once, don't
bother with realistic depth at all. To describe the
sound as more “alive” might seem strange, but
in defense I must say finally on this subject of
realism that there is an ineffable quality about
any behaviourally designed sounds, regardless of
whether the behaviour makes sense. Although I
cannot yet formalise this in my own words it is
illustrated by some anecdotes and quotes and is
certainly worth further study with more formal
experiments. In one observation, students playing
with an explosion generator and a billiard ball col-
lision simulator simply did not want to stop. They
knew instinctively the difference between this and
a system that was just playing back a wide choice
of samples. (Perhaps this can be accounted for
by novelty against an generational experience of
samples). Without any visual accompaniment the