Physical Modelling for Sound Synthesis - Game Sound Technology and Player Interaction: Concepts and Developments - page 348

to model this phenomenon for rough surfaces.
(p. 214)

the name may suggest, this method uses modal
synthesis and the intention is that the modal data
has a physical meaning. The framework suggests
methods for determining the modal frequencies
from the recording of a real sounding instrument,
for example “peak-picking from spectra calculated
by Fourier analysis” (Cook, 1997, p. 39). Although
automation can help in setting the modal data,
the author suggests “it relies heavily on human
analysis and decisions” (p. 39). To create an ap-
propriate excitation signal, the suggestion is made
to record an actual object striking a non-resonant
surface. So, if one wished, for example, to simu-
late a virtual object being struck by a mallet, one
should use the recording of a real mallet striking
a non-resonant body. If the body being struck is
completely rigid and non-resonant then the sound
made will be due only to the mallet and should
therefore capture the excitation force, which can
be used as an excitation signal in the virtual world.
The second algorithm is called Physically
Informed Stochastic Event Modeling (PhISEM).
This involves “the overlapping and adding of small
grains of sound” (Cook, 1997, p. 40), a familiar
process known as granular synthesis. The method
is motivated by instruments like the maracas or
tambourine, the characteristic sounds of which
are made up of many short, discrete sound events.
While these instruments may be uncommon in
the average computer game environment, Cook
notes that PhISEM algorithms can also be used to
synthesize many everyday sounds like footsteps
on gravel or dripping water. The process involves
analyzing a sound-producing system and possibly
the waveform of the sound it produces. From this
analysis, rules are created to map the variables of
the system, such as the shake velocity of a maracas,
to the parameters of a granular synthesis algorithm.
One of many contributions in the area by Kees
van den Doel and Dinish K. Pai was published
in 1998. The Sounds of Physical Shapes (Doel &
Pai, 1998) documented the creation of their Sonic
Explorer application which adds object contact
sounds to real-time simulation environments.

While its implementation is not presented in
their paper, the idea of using 1 / f noise for a scarp-
ing sound has been utilized to good effect in more
recent projects.
Around the same time as the above work
appeared, William W. Gaver published some
research on auditory event perception (Gaver,
1993b) and synthesizing sounds for auditory
icons (Gaver, 1993a). In the former work, Gaver
notes that “sound provides information about an
interaction of materials at a location in an envi-
ronment ” (Gaver, 1993b, p. 5) and he explores
“everyday listening” as the process of listening
to events rather than sounds. From this, a method
for describing sounds by attributes of the sound
source is proposed. In the latter work, Gaver
states that auditory icons can add functionality
to a computer environment and, while the digital
samples cannot be easily manipulated, it is possible
to synthesize them. He then describes algorithms
to synthesize many everyday sounds as specified
by parameters that describe the sound producing
event, unlike many synthesis methods at the time
which synthesized sound in musical terms. This
was an important step towards synthesizing object
contact sounds based on their physical attributes, a
necessity of physically motivated sound synthesis
in virtual environments.
A few years later, in 1997, Perry Cook pub-
lished research carried out on physical modelling.
Physically Informed Sonic Modeling (PhISM):
Synthesis of Percussive Sounds appeared in
the Computer Music Journal (Cook, 1997) and
documented Cook's work in creating a frame-
work which could synthesize the sounds of a
wide range of percussion instruments. Here, two
synthesis algorithms are presented motivated
by two distinct types of percussive instruments.
The first, Physically Informed Spectral Additive
Modeling (PhISAM), deals with resonant percus-
sive instruments like a marimba or cowbell. As