Information Technology Reference
In-Depth Information
sounds alone, like a musical instrument, seemed
engaging enough to just keep pressing the buttons
(in a way that nobody would ever do with a sample
replay). One described it as “satisfying”, another
as “moreish” (addictive). Another, more academic
explained that the continual subtle difference
was what forged its identity and begged further
exploration (I suppose in the spirit of Deleuze) as
opposed to the anaesthetic, throw away emptiness
of repetitious samples. One emotional testimony
stays in my mind from an online discussion in
2005 about synthetic game sound. It remains
inspiring to my work even today. Cosinusoidal
(screen name) writes:
priori
, through visual priming, by experience,
or on a short timescale by the “rules of auditory
scene analysis” (McAdams & Bigand, 1992)
which amount to an innate understanding of phys-
ics (and by Chion's (1994) rules of audio-vision
synchronisation). I like to extend this idea to
encompass “behavioural plausibility” in general.
As an example, an innate physical behaviour is
that all systems are in energetic decay unless a
new source of energy is supplied. A bouncing
ball must decay (decreasing period between col-
lisions), because if it makes sound there must be
loss, ergo less kinetic energy, and thus shorter
subsequent bounces. We know this at a deep level,
learned through exposure to a lifetime of physics,
or perhaps the result of inherited structure (in the
sense of Chomsky's innate grammar disposition).
A ball with increasing energy (apparent through
increasing collision intervals, perhaps because a
force is silently being applied to it by a basketball
player) seems to be playing backwards, even
though the energetic decay of individual collisions
is correct. Thus the larger scale behavioural feature
dominates the smaller scale one. This interpreta-
tion of energetic growth, decay or maintenance is
given by Pierre Schaeffer (discussed by Miranda,
2002, p. 127).
Where is the limit of this behavioural realism?
At what point does real become real enough?
The answer, in our context, is
where it serves
its purpose for arts and entertainment
. Once the
depth of object behaviour is good enough we need
go no further. Using super-computers and many
hours of computation to produce sounds, misses
a crucial point. There is a middle ground between
improving upon sampled sounds and a scientific
simulation on which lives depend. Failing to
distinguish these goals and chasing brute force
implementations with computationally precise,
but sensibly inaccurate, or perceptually irrelevant
results is a mistake in my opinion (unless you are
actually performing simulations for civil aircraft
noise abatement purposes). Knowing which are
the relevant parameters is what the art of practical
“Real synthesis in computer games is something
I imagine every night to help me drift off into
dreams/sleep . I remember being 7 years old and
playing C64 computer games and the fact that
the computer was some how alive making these
sounds brought tears to our eyes . Hopefully games
developers will realise that real synthesis is pro-
foundly more immersive than samples could ever
be, today reading your views on the same thing
has gave me hope [sic] that computer games might
one day return to synthesis methods.”
Plausibility and Edge cases
A sonic object
captures
some feature of a real
source when one or more parameters can be
ascribed corresponding to a real physical vari-
able. For example, the order of a low pass filter
combined with the delay time of a ground image
crudely captures the distance of an aircraft where
height and all other variables remain fixed. It
sounds real because it directly matches something
that happens in reality (multiple path destructive
interference causing a comb filter sweep).
Beck's (2000) “
acoustic viability
” captures
the behaviour of a synthetic musical instrument
once the parameter space accords with our sensory
expectations. These expectations can be set up
a
Search WWH ::
Custom Search