Content uploaded by Ulf A. S. Holbrook
Author content
All content in this area was uploaded by Ulf A. S. Holbrook on Nov 07, 2020
Content may be subject to copyright.
An approach to stochastic spatialisation
A case of “Hot Pocket”
Ulf A. S. Holbrook
RITMO Center for Interdisciplinary Studies of Rhythm, Time and Motion
Department of Musicology
University of Oslo
u.a.s.holbrook@imv.uio.no
ABSTRACT
Many common and popular sound spatialisation techniques
and methods rely on listeners being positioned in a “sweet-
spot” for an optimal listening position in a circle of speak-
ers. This paper discusses a stochastic spatialisation method
and its first iteration as implemented for the exhibition Hot
Pocket at The Museum of Contemporary Art in Oslo in
2017. This method is implemented in Max and offers a
matrix-based amplitude panning methodology which can
provide a flexible means for the spatialialisation of sounds.
Author Keywords
Sound spatialisation, sound art, sound for exhibitions, im-
mersive audio, sound and music computing.
CCS Concepts
•Applied computing →Media arts; Sound and mu-
sic computing;
1. INTRODUCTION
This paper covers the implementation of a stochastic spa-
tialisation technique using random walks for the exhibi-
tion Hot Pocket at The Museum of Contemporary Art in
2017.1The backgrounds for the technique, implementation
in the space and considerations in terms of working with
spatial sound design for exhibition works will be discussed.
Hot Pocket was an exhibition by the Norwegian artist Tori
Wr˚anes2which incorporated video, sculpture, sound, per-
formance and documentation of previous performances as a
gesamtkunstwerk. The visitor was invited into a completely
transformed space where all walls and floors were covered
in carpets giving the impression of being in a cave or en-
capsulated inside the belly of a large animal. The sound
implementation had as its objective to completely fill the
space and to envelop the audience to heighten the feeling of
navigating this indefinable interior.
2. BACKGROUND
1The Museum of Contemporary Art is part of The National
Museum: http://www.nasjonalmuseet.no. The exhibition
ran from April 21st-September 3rd 2017.
2http://www.toriwraanes.com
Licensed under a Creative Commons Attribution
4.0 International License (CC BY 4.0). Copyright
remains with the author(s).
NIME’18, June 3-6, 2018, Blacksburg, Virginia, USA.
There are no lack of spatialisation technologies currently
available, which highlights this as a very fertile and popu-
lar field of research. The availability of high-density loud-
speaker arrays (HDLA) is still small but the advances be-
ing made, could help the growth of spatial composition
instead of the more traditional timbral composition with
stereo diffusion over a speaker array [4]. Several existing
HDLA spaces are used for a range of performances, where
the audience would generally be seated when experiencing
the works. In many cases multi-channel setups for exhi-
bition venues will generally be designed and setup on an
ad-hoc basis, which does not render it as flexible as venues
with an established infrastructure for multi-channel sound
reproduction, yet on the other side it can facilitate a think-
ing “outside the circle.”
Figure 1: View of Banksalen at The Museum
of Contemporary Art. Photo: Annar Bjør-
gli/Nasjonalmuseet.
For a good overview of technological developments of spa-
tialisation methods see [6]. Several existing algorithms were
considered for this project, among them Vector-Based Am-
plitude Panning (VBAP) [7], Distance-Based Amplitude
Panning (DBAP) [2], flocking or swarming behavior [8], am-
bisonics [5] and Wave Field Synthesis (WFS) [1]. Although
these are all spatialisation algorithms with strong merits for
their uses, we found that our rather difficult wishes of per-
petually traveling sounds would be difficult to achieve with
these methods, specifically when considering the space it
would be implemented in. The resulting method was heav-
ily inspired by these existing algorithms.
3. IMPLEMENTATION
The project was implemented in Max3. A multichannel
soundtrack could easily been assembled in a DAW like Reaper4
but the same spatial distribution would be heard every day
of the exhibition period and our aim was to create a dis-
3https://cycling74.com
4https://www.reaper.fm
tribution of sound which create an ever-changing sonic ap-
pearance of the exhibition, with varying densities of sound.
Inspired by Iannis Xenakis’ (1922-2001) approach to dy-
namic stochastic synthesis [10], the data for spatial distribu-
tion was generated using random walks. The random walk
paradigm allows for higher or lesser degrees of control, de-
termined by how“random” one wants the distribution to ap-
pear: “The fluctuation speed of a parameter is directly pro-
portional to the step size of its random walks: the smaller
the steps, the slower the rate of change in that parameter”[3]
p. 79. The audience would generally not be seated, which
would not be a well-suited situation for a fixed sweet-spot
reproduction. With constant motion of the sound trajecto-
ries, the aim was that the audience experience the sound as
immersive and continuously moving. First, the number of
files to be played back was determined, then a mono sound
stream is panned around the rooms based on the stochastic
data. The data is generated for three parameters of ampli-
tude, speaker number and transition time. The transition
time is the cross-fade time between two speakers, where
the amplitude value states to/from values of speaker transi-
tion. The transition time also defined the rate of reading the
data which in effect means that the sound never “lingers”
at any one speaker. At startup, the number of sound files
are selected, the initial matrix data is generated and the
internal clock controlling triggers is started. The stochastic
data used to spatialise the sound is regenerated every time
a new group of files are selected. Although of equal lengths,
each file will end at different times due to the transition
time parameter. When all files have stopped playing, a new
selection of files and new data is generated.
The stochastic technique has several advantages among
them being easily scalable data ranges, it does not rely on a
sweet-spot position and it makes no assumptions about ei-
ther speaker nor audience placement. By adjustment of step
sizes for all three main parameters, a large variety of tra-
jectories can be achieved - from the seemingly static to the
random. The disadvantages are considerable lack of control
over the spatial distribution of sound in a given space and
adapting a work for different spaces is more time consuming
than using Ambisonic encoded audio.
3.1 Equipment
The system was designed for an arbitrary amount of speak-
ers and in total we ended up using 35 speakers and 4 sub-
woofers produced by Genelec, in a range of different models
(6010, 8010, 8020, 8030, 8330, 7050), distributed as evenly
as possible around the space. The difference in frequency
response and power from the smaller to larger speakers was
considerable but seen as an advantage. A Mac mini run-
ning Max along with two MOTU 24Ao soundcards5, each
providing 24 output channels over D-Sub to XLR. Where
possible the speakers were hung at 3 meters above the floor,
angled down 45 degrees.
4. NON-SPECIFIC SPATIALISATION
The stochastic methodology arose from a desire to take ad-
vantage of the inherent idiosyncrasies and limitations of the
space. The Museum of Contemporary Art was housed in
a former bank, which adds to the rather overwhelming ap-
pearance of the space and consists of rooms with dramati-
cally varying sizes and ceiling heights. The artist’s aim was
to transform the space completely, which was achieved in
large parts by the sound, the lighting and the carpets cov-
ering the walls and floors. The notion of non-specific spa-
tialisation [9] emphasises that this methodology does not
5http://motu.com/products/avb/24ai-24ao
seek to ’simulate precise locations or directions of (usually
point) sources’ [9] p. 241. As such, in many situations one
aims and wishes to recreate a specific real-world sound scene
or to synthesise a ’new’ space to create an imaginary but
“real” space. The aim for Hot Pocket was not to attempt
to create a believable real space but to emphasise the ex-
perience of being immersed in a cave or a large animal of
some sort.Non-specific spatialisation is helpful when consid-
ering the lack of standarization among the various setups for
multi-channel sound reproduction found in the world. But
specifically when attempting to design a holistic and func-
tioning spatialisation system for highly irregular spaces.
5. CONCLUSION AND FUTURE WORK
Considering the site and the movement of people through
the site, the decision to prototype a stochastic mode of spa-
tialisation was deemed successful both by the artist and by
the author. For setup, experimentation and changes, the
method proved very flexible, where only simple step-size
parameters needed to be changed to achieve a very differ-
ent spatialisation in the space. The abstractions from this
prototype implementation and the tools currently being de-
veloped are thought to exist at the very base of a composi-
tional or sound art process allowing for a direct and imme-
diate tool for exploring spatial organisation and placement.
The advantage of this process is its flexibility as it is de-
signed for used with an arbitrary amount of speakers and
is not reliant on careful circular (or near-circular) speaker
arrangements.
6. ACKNOWLEDGMENTS
I would like especially to thank Tori Wr˚anes and Rune An-
dreassen, along with his whole team.
7. REFERENCES
[1] A. J. Berkhout. A holographic approach to acoustic
control. Journal of the audio engineering society,
36(12):977–995, 1988.
[2] T. Lossius, P. Baltazar, and T. de la Hogue.
Dbap–distance-based amplitude panning. In ICMC,
2009.
[3] S. Luque. The stochastic synthesis of iannis xenakis.
Leonardo Music Journal, 19:77–84, 2009.
[4] E. Lyon. The future of spatial computer music. In
ICMC, 2014.
[5] D. G. Malham and A. Myatt. 3-d sound spatialization
using ambisonic techniques. Computer music journal,
19(4):58–70, 1995.
[6] N. Peters. Sweet [Re] production: Developing sound
spatialization tools for musical applications with
emphasis on sweet spot and off-center perception.
McGill University, 2011.
[7] V. Pulkki. Virtual sound source positioning using
vector base amplitude panning. Journal of the audio
engineering society, 45(6):456–466, 1997.
[8] S. Wilson. Spatial swarm granulation. In Proceedings
of the 2008 international computer music conference.
SARC, ICMA, 2008.
[9] S. Wilson and J. Harrison. Rethinking the beast:
Recent developments in multichannel composition at
birmingham electroacoustic sound theatre. Organised
Sound, 15(3):239–250, 2010.
[10] I. Xenakis. Formalized music: thought and
mathematics in composition. Number 6. Pendragon
Press, 1992.
