the
music
we
decided that, given the limitations of the room size and computing equipment
available, we would set the PowerBook to play a continuous wash, which
would create a continuum over which the files triggered by the x10s
would be played via the PC.
the
wash was composed from a recording of John playing guitar, distorted
through a Marshall amp, and from there imported into ProTools Free,
where two channel reverb was added. its melodic nature was simple and
repetetive, forming a base for the less regular sounds to rest upon.
Miriam then created graphics in Photoshop 6.0 which were imported into
Flash, based on a colorisation and deconstruction of a photograph of
a begonia leaf. Flash is normally used for vector graphics animation,
but by overlapping bitmaps in different colours on several layers and
fading their alpha channels in and out in different phases, an effect
of a bitmap slowly shifting in colour could be acheived.
the
other samples which were used to compose the seven shorter segments
were quite diverse in nature. this was intentional - haiku poets do
not restrict themselves to the contemplations of nature for which the
style is best known, in fact many haiku poems can contain intense and
very moving emotions within their essential framework. we collected
a wide variety of samples; beginning with a series of phrases composed
using soundfonts in Linux, some recordings of flowing water, raking
stones and childrens voices from a neighbouring park, Miriam singing
vocal harmonics, and an excerpt from Josquin's "Missa Pange Lingua",
arguably one of the most purely beautiful pieces in Western music history.
the
samples were then processed in ProTools Free and Cool Edit Pro in Windows98
to create pitch shifts and filtering of a spectrum of harmonics using
parametric eq. a tonal centre around A had already been established
in the soundfont samples and the background wash, with some shifts to
relative centres of C and F. a focus on the harmonic spectrum of these
sounds invited a contemplation of their inner tonal relationships and
of the nature of sound.
further
processing was then done in Linux, taking this premise further into
deconstructing the harmonics within the samples. the NoTAM suite of
FFT-based granular synthesis applications were excellent for this purpose.
Ceres3 produces some extremely interesting
effects with sieving and comb filters, and also allows visual cut and
paste of the soundfile analysis. Mammut's
most striking features are the Block Swap and Phase Shift options, which
chop up the soundfile to produce a longer file created from small samples
- this is probably the effect for which granular synthesis is best known.
inputting a negative number in Phase Shift creates a soundfile formed
from reversed segments of the original. knowing a few 'secret combinations'
we produced some fairly complete compositions simply from processing
the files.
some
files were then arranged in ProTools Free, while others were left raw;
the arranged files were then combined with Flash graphics, this time
using more specific photographs of a snail, a fish, waves and some stones
to complement the less processed samples, with some deconstruction applied
in the same manner as with the begonia leaf. in selecting an appropriate
combination of files and conditional scripting, visitors to the installation
could then compose their own experience according to which motion sensors
they chose to trigger, and as they spent more time with the piece, they
could build up selections of complementary - or disjunct - files.
the
installation - goldsmiths college computer music conference
the
first incarnation of < .h> was installed in the Electronic Music
Studio at Goldsmiths College for the Computer Music Conference, Feb
24, 2001. unfortunately there were a few obstacles in the process of
arranging the installation which meant that we had little time to test
the sensetivity and timing of the x10 controllers before the performance.
we set the seven shorter files to be triggered by the PC upon receiving
signals from the seven individual motion sensors, and then experimented
with the placement of which file was to be triggered by which motion
sensor in order to allow various structures to arise within the piece.
we were particularly concerned about the seventh sensor, which was placed
below the fishbowl - the remaining six were hung on aluminium rods from
the ceiling in an even distribution. as the fishbowl could not be seen
on entering - it was hidden behind boxes so that it could be discovered
as visitors explored the room - this was logically the climax of the
piece, and so its soundfile needed to be of significance. this also
meant that the soundfiles around it and their graphics needed to create
a smooth transition to this moment. gentler, less conspicuous files
could then be placed near the doors to introduce the piece.
it
was becoming clear that too many graphically accompanied soundfiles
were overwhelming both to the composition and in their drain on CPU
power. we filtered them out by restricting the piece to three sound-only
.wav samples, with the remaining four sensors triggering graphics plus
soundfiles in .swf format.
with
some fine tuning the piece arrived at greater stablity. some crashes
occured in the initial stages (including some hilarious moments where
the projector attached to the PC blared out a classic Windows "Blue
Screen of Death"). this interruptions were very much unwanted!
- and so we set some conditions upon the soundfiles - certain files
would only play if others were not playing, and each motion sensor was
set not to re-trigger within one minute. we found that the motion sensors
were extremely sensetive, and while this created many problems to begin
with, it also lead to some very beautiful moments where the piece virtually
played itself - sometimes the motion of Basho the fish within the fishbowl
was enough to trigger a sensor, and certain sensors would regularly
be triggered by the heat of the projectors turning on and off. this
randomness would ensure that it was not obvious to the visitor which
of their motions were triggering which pieces, avoiding an over-simplistic
effect.
this
was very much the first incarnation of < .h>, a piece which can
take many forms - this was probably the smallest scale which we will
work with, but it is possible to extend it to a cluster of larger forms,
and to produce the work in a sound-only format if projectors are not
available. the form of the installation is ultimately determined by
the space in which it is installed, bringing visitors from that space
into the stillness of a world which, while not entirely seperate from
their own, draws one into an experience of contemplation and inner peace.
<MR>
|
the
future
While
windows was almost adequate for the job, there were problems with
latencies (partly windows and possibly partly anti-virus software)
and also an annoying buildup of expired player instances that
couldn't be cleaned up automagically. Linux with low latency patches
and some scripting should improve that situation drastically.
One
thing about the Goldsmith's installation was that it couldn't
be left unattended. A design goal for future versions is that
the thing can sit there for days happily churning out sounds and
video without the need for constant house cleaning... or any house
cleaning at all.
At
the moment we're talking to various people about future versions
and the chances are good it will be appearing again quite soon
in a reasonably cool commercial setting. More news as it happens!
<JL>
|
©2001
Miriam Rainsford and John
Littler.
For
further information visit:
www.h-project.n3.net
www.moshi-d.com