Content uploaded by Stephen Barrass
Author content
All content in this area was uploaded by Stephen Barrass on Sep 18, 2014
Content may be subject to copyright.
SoniHED – Conferen ce on So nificatio n of Health and Environmental Data 12 September York, UK
ACOUSTIC SONIFICATION OF BLOOD PRESSURE
IN THE FORM OF A SINGING BOWL
Stephen Barrass
Digital Design and Media Arts,
University of Canberra,
Canberra, Australia
Stephen.Barrass@canberra.edu.au
ABSTRACT
The Hypertension Singing Bowl is an Acoustic Sonification
shaped by a year of blood pressure data that has been 3D printed
in stainless steel so that it rings. The design of the bowl was a
response to a medical diagnosis of hypertension that required
regular self-tracking of blood pressure. The culture of self-
tracking, known as the Qu antified Self movement, has the motto
“self knowledge through numbers”. This pap er de scribes the
process of designing and digitally fabricating a singing bowl
shaped from this b lood pre ssure data. An iterative design
research method is us ed to identify import ant stages of the
process that include the choice of a sonic metaphor, the
prototyping of a CAD baseline, the mapping of data to shape,
and the acoustics of the mapping. The resulting Hypertension
singing bowl is a m editative contemplation on the dataset that i s
a reminder to live a healthy lifestyle, and a poetic alternative to
generic graphic plots of the Quantified Self.
1. INTRODUC TION
The increasing availability of low cost wearable sensor products
has led to a growing interest in self-tracking in sports, health and
fitness. The Quantified Self (QS) movement advocates “self-
knowledge through numbers” [1] through the analy sis of this
data. Participants in QS meet-ups are invited to “share what you
are doing, and learn from others” by showing visualizations,
and telling stories about datasets, self improvements,
technologies and other aspects o f self -tracking culture [2].
Mark Carrigan observes that stories about self-
tracking often in clude p ersonal context and a qualitative
interpretation of the numbers. He introduced the term “qualified
self” to refer to “self-knowledge through words” that comes
from telling stories about the data. He goes on to define the term
“qualitative self-tracking” as “using mobile technology to
recurren tly record qualities of experience or environment, as
well as reflections upon them, with the intention of archiving
aspects of personal life that would otherwise be lost, in a way
susceptible to future review and revision of concerns,
commitm ents and practices in light of such a review” [3].
Jenny Davis m akes the point that telling stories about
self-tracking data can also be a mechanism for constructing self-
identity.“Self-quantifiers don’t just use data to learn about
themselves, but rather, use data to construct the stories that they
tell themselves about themselves” [4]. She also observes that
personal reflections on the data can go beyond words to include
artistic constructions such as a po em, or a collage. D eborah
Lupton expands on the kinds of data that are collected by self-
trackers in her analy sis of cultures of self-reflexion. “Many self-
trackers record non-quantifiable data as part of their practice,
including journaling accounts of their daily activities, emotional
states and relationships, collecting audio data or visual images
and producing visualisations that centre on their aesthetic or
explanatory properties rather than their representation of
numbers. [5]
In a recent post on the Quan tified Self site, Enrico
Remirez showed images of physical visualizations that included
a 3D bar chart made fro m children ’s playing blo cks, and a
sculpture made from graphs cut out of cardboard and bound
around a spine [6]. Physical visu alizations like these are usually
considered to be educational props, or artistic interpretations of
the data. How ever, a recent study by Yvonne Jansen and
colleagues found that a 3D print of a 3D dataset can be more
effective for 3D information retrieval tasks than a screen-based
version [7]. In another study, Rajit Khot and colleagues found
that parti cipants w ere mor e conscious of their daily physical
activity when heart rat e dat a was presented as a 3D printed
object than when it was shown on a screen [8].
These studies support the proposal in this paper that
stories about self-tracking may not necessarily have to be told in
words to enable the personal reflection that may transform
numbers into identity. Stories can be told non-verbally through
paintings, sculptures, and music. Stories about numbers may be
told non-verbally through graphic visualizations, physical
visualizations, and data sonifications. Building on these
techniques, this paper introduces a new technique, known as
acoustic sonification, as a medium for telling stories about
numbers. Acoustic sonifications are physical visu alizations that
also make sounds [9]. Could an acoustic sonific ation be
constructed from self-tracking data? Would this sonic object
also facilitate story telling and promote reflection on personal
health and fitness? Could the sound increase the curiosity to
explore, or enable alternative perceptions and interpretations of
the dataset?
These questions motivated the experiments described
in the rest of this p aper. The background section presents a brief
history of acoustic sonification, along with some early
examples. The b ody of the paper describ es the design and
realization of a prototype of an acoustic sonification designed
for a dataset consisting of blood pressure readings taken over a
one-year period. The discussion reflects on the experiment in the
context of the questions raised by theories of the quantified and
SoniHED – Conferen ce on So nificatio n of Health and Environmental Data 12 September York, UK
qualified self. The paper concludes with a summary of the
process of designing an acoustic sonification that includes stages
for furth er research and development.
2. BACKGROUND
In 2009 a CAD model of a whistle was uploaded to the
Thingiverse.com 3D printing community site. The whistle
generated considerable attention, because it did something no
other 3D printed object had done before, it produced a sound.
But what generated most attention was the difficulty o f 3D
printing a version that actually whistled. The variability in the
results p roduced by different printers and different settings
highlighted the intimacy of the coupling between shape,
material and sound.
In 2011 Arvid Jense documented 40 experiments with
the 3D printing of CAD designed musical instruments. The
experiments included whistles, blown tubes (e.g. pan pipes),
Helmholz resonators (e.g. a blown bottle), percussive temple
blocks, and “impossible” inst ruments of a complexity that is
made possible with digital fabrication processes [10]. Most of
the instruments did not produce any sound at all, and Jense
observed that the precision of edges, angles, holes, and surfaces
was critical. He also noted that instruments printed in plastic
did not generally produce sounds of a musical quality, with the
exception of one particular t emple b lock that had an infill
pattern that produced a more wood like timbre.
In 2013 the Stanford University Centre for Computer
Research into Music and Acoustics (CCRMA) organised a
workshop titled 3D printing for Acoustics to introduce product
designers to 3D printing with “music making in mind”. The
participants modelled acoustic objects with Computer Aided
Design tools, parametric equations, and 3d scans of pre-existing
objects, to produce a slide flute, a pretzel shaped flute, and a
percussive washboard [11].
In 2013 the online 3D printing service,
Shapeways.com, announced the dawning of a “New B ronze
Age” with the introduction of the capability to 3D print CAD
models in bronze and brass. Online 3D printing services, like
Shapeways, provide access to the latest developments in digital
fabrication technologies that can print a growing range of
materials. The development of 3D printing in ceramics and
glass has been driven by home-wares, and jewellery is driving
printing in stainle ss steel, brass, bronze, silver, and gold. The
range of materials continues to expand, and there are almost
daily announcements of new printers capable of fabricating
rubber, concrete, carbon fibre, bone structures, arteries, organs
and even fo od. The size of the objects is increasing, and there
are now even 3D printers at an arch itectural scale. High
resolution printers can produce mechanisms wi th moving parts,
and the capability to print in multiple materials allows
electron ic circuitry to be embedded. Examples of 3D printed
acoustics include gramophone records [12], speakers [13],
music boxes [14], and noise mufflers [15]. Researchers in the
Creative Machines Lab at Cornell University recently 3D
printed a fully functioning loudspeaker with plastic, conductive
and magnetic parts [16].
The discovery of the resonant properties of metals in
the Bronze Age let to the invention of instruments such as
gongs, bells and bowls. The musical properties of brass makes
it the material of choice for tubas, horns, trombones, trumpets
and other instruments. The capability to 3D p rint in these
metals expands the range of potential 3D printed instruments.
In 2011, I 3D printed a bell in stainless steel to test that it would
ring, which it did [9]. The modulation of the shape of the bell
by a digital dataset caused it to ring with a different pitch and
timbre, and the effect of the dataset on th e acousti cs of the b ell
was visible in a spectral analysis [9]. This experiment
supported the hypothesis that info rmation about a dataset could
be heard in an acoustic sonification. However, it also raised
many questions. What effects do different mappings of the data
onto the shape have on the acoustics? What is the relationship
between physical acoustics, and the auditory perception of
informative relations in the data? What kinds of information
can be understood from different mappings of data into shape
and acoustics? What other shapes beside a bell could be used in
acoustic sonification? What effects do other instrument shapes
have on the interpretation of meaning from the sounds of
interacting with the object?
3. HYPERTENSION
Hypertension, or high blood pressure, is a common disorder of
the circulatory system, affecting around one in seven adult
Australians. It is also known as “the silent killer” because there
are no symptoms, and many people are unaware that they have
this potentially lethal condition. Experts recommend that
everyone should have their blood pressure checked r egularly.
A medical diagnosis of hypertension led me to begin
self-tracking my blood pressure with a cuff that sends the
readings to an App on a mobile phone. The cuff measures
systolic pressure, which is the maximum pressure on the arteries
when the heart beats, and diastolic pressure, which is the
minimum pressure when the heart relaxes. This data is typically
plotted in a time series graph, as shown in the screenshot from
the App in Figure 1.
Figure 1. Plot of blood pressure readings
4. DESIGN PROCESS
The plots of my blood pressure are generic and could have been
share prices o r CO2 emissions. Surely there was a way to make
this personal dataset more personal, and more engaging. Could
an acoustic sonification provide an antidote to the silent killer?
SoniHED – Conferen ce on So nificatio n of Health and Environmental Data 12 September York, UK
In previous exper iments the choice of a bell provided
a metaphor for interaction, and set up the expectations of how
the sonification should sound. Other 3D printed instruments
could also be sonic metaphors e.g. the pan-pipe, whistle, flute
percussion block, washboard, rattle and gramophone record.
However, after some consideration, I selected the Tibetan
Singing Bowl, because it is associated with meditation and other
relaxation therapies that can lower blood pressure. Antique
singing bowls are sought out for their unique sounds which are
the result of hand crafting from alloys that include gold, silver,
mercury, copper, iron, tin, lead, zinc, nickel and other trace
elements. Today singing bowls manufactured by casting in
bronze are more uniform in shape, material and the sounds they
produce. The modulation of the shape of a 3D printed singing
bowl by a personal health dataset might also reintroduce a
unique sonic character to each bowl.
The simple shape of a singing bowl, shown in Figure
2, makes it straight forward to model as a CAD mesh as shown
in Figure 3.
Figure 2. Tibetan singing bowl.
The mesh was constructed from 3D graphic primitives in the
Processing 3D graphics programming environment [17].
Figure 3. CAD mesh of a Tibetan singing bowl
The next stage was to map blood pressure data onto
the CAD mesh. In the previous experim ent polar HRTF data
was mapped in a circle around the bell shape. The profile of the
bell wall was then modulated by the HRTF parameters at that
angle. The mapping from data to shape was info rmed by a study
of the relationship between the acoustics and shape of bells by
Neale McLachlan, who found that wall thickness and profile
affect pitch and ti mbre [18].
The blood pressure dataset consists of pairs of
systolic/diastolic measurem ents reco rded over the period of a
year. This dataset does not have a polar spatial dimension that
maps directly to the circular shape of the bowl. As a first
experiment, the time axis was mapped rad ially outward from the
top centre of the bowl to th e outer edge. The circumferen ce of
the outer wall was modulated with the systolic data, and the
circumference of the inner wall with the diastolic data. The
minimum thickness for 3D printing in stainless steel is 1.5mm.
The modulations of thickness were added to this core, resulting
in wall s up to 5mm in thickness, as shown in the CAD model in
Figure 4.
Figure 4. Blood Pressure Singing Bowl 0.0 – CAD mesh.
The size of the CAD mesh was reduced by removing
overlapping vertices with Meshlab [19]. The mesh was then
checked for holes and repai red with netfabb [20]. The cleaned
mesh was then uploaded to Shapeways and 3D printed in
stainless steel. The resulting bowl, shown in Figure 5, is 64mm
in diameter, with volume 37.3 cm3, and w eighs 275g.
Figure 5. BP Singing Bowl 0.0, 3D printed in stainless steel.
Striking the side of the bowl with the Puja stick
produces a ringing tone at 3628 Hz that lasts for 3 seconds.
Rubbing the stick around the rim produces a metallic sound but
the bowl does not reson ate and sing like a traditional bowl. The
failure of this first experiment to produce a bowl that could sing
led to the re-examination of a traditional bowl which had w alls
that were only 2mm thick, and the observation that the walls
were twice as thick. Upon reflection on this process, it would
SoniHED – Conferen ce on So nificatio n of Health and Environmental Data 12 September York, UK
have been more efficient to begin by 3D printing baseline bowl
as a test before moving on to the m apping stag e.
This observation led to an iteration in the design of the
form with the specification that the thickness should be 2mm.
This constraint required a redesign of the mapping of the data
onto the shape. Rather than mapping the timeline along a radius,
it was mapped around the circumference. The pairs of
systolic/diastolic data were assigned to radial spokes that
connect the rim to the base, as shown in Figure 6. The systolic
pressure moderates the radius of the upper half of the spoke, and
the diastolic data moderates the radius of the lower half.
Variations in the data move the upper and lower parts of each
spoke inward and outward to produce an individual acoustic
effect at each spoke. In theory , rubbing the rim with the stick
should activate the spokes to addi tively synthesise an acoustic
sonification of the entir e dataset.
Figure 6. BP Singing Bowl 1.0 – CAD mesh.
As before, the CA D mesh was reduced, repaired and
uploaded to be 3D printed in stainless steel. The resulting bowl,
shown in Figure 7, is 100mm in diameter, with volume 18.7
cm3, and weighs 162g.
Figure 7. BP Singing Bowl 1.0, 3D printed in stainless steel
Striking the side of the bowl with the stick produces a
dominant partial at 609 Hz that rings for 10 seconds. The
ringing tail has a tremolo effect at 2Hz visible in the waveform
in Figure 8.
Figure 8. Audio waveform of BP Singing Bowl 1.0 when struck.
As well as the bell tone, the strike also produces an
unusual hissing sound that lasts 2-3 seconds, as can be seen as a
grey band between 1500 and 2000 Hz in the spectrogram in
Figure 9. The tremolo and hissing effects, which are not heard
in a tradi tional bowl, m ay be caused by the data spokes.
Figure 9. Audio spectrum of BP Singing Bowl 1.0 when struck.
When the rim was rubbed with the stick the bowl
began to hum, and then sing like a traditional bowl. The audio
waveform in Figure 10. shows the bowl continues ringing for 16
seconds after rubbing ceases.
Figure 10. Audio waveform of BP Singing Bowl v1.0 when
rubbed and left to ring.
The frequency analysis in Figure 11. shows
broadband low frequencies from the rubbing motion, a
SoniHED – Conferen ce on So nificatio n of Health and Environmental Data 12 September York, UK
dominant partial at 609 Hz, and higher partials due to other
resonances that may include the data spokes.
Figure 11. Frequency analysis of BP Singing Bowl v1.0 when
rubbed and ringing.
5. DISCUSSION
The singing bowl provides a metaphor for interaction and
engagem ent w ith the data embedded in i ts sh ape. The tangibility
of the object invites handling, and the sounds that it produces
spark curiosity to explo re it further. The bowl can be tapped, or
rubbed with different rates and forces to produce different
sounds, and could even be used in a musical performance. The
association with m editation and relaxation therapies contribute s
to a narrative of contemplation and reflection on the dataset as a
means of self-discovery and self-improvement.
The non numeric and non verbal nature of the singing
bowl raises the question of whether someone could really
understand information about a dataset from this object. Data
visualisation theorist Jaques Bertin defined structural
interrelationships that emerge from a dataset as a whole as a
higher level of information than the data values in isolation [21].
From this perspective, the capability to listen to the way the
entire dataset affects the sound, rather th an listening to
individual points, could provide an understanding of higher-
level structure. However, there is much more work required to
understand the perceptions of a dataset that can be obtained by
interacting with an acoustic sonification. An initial step would
be to 3D print a “baseline” bowl that does not have a dataset
embedded in it. The acoustics of this baselin e could be
compared wi th b owls constructed from datasets that vary in
systematic ways. It should be noted that sonification depends
critically on a human listen er. Perceptu ally based evaluation will
involve listening for specific features in a constructed dataset.
User-centred evaluations will involve testing the usefulness of
the acoustic sonification in specific tasks.
6. CONCLUSION
These experiments with the acoustic sonification of blood
pressure data have identified important stages of the design
process that can guide future designs and further research :
1. Sonic Metaphor: guide interaction, establish sonic
expectations, provide a context for interpretation.
2. Baseline Prototype: a CAD model of the Sonic
Metaphor, which is 3D printed to test that it works and
makes a sound.
3. Data to Shape Mapping: the mapping of data axes
onto geometric axes of the Sonic Metaphor, and data
values onto geometric variations of the shape.
4. Acoustic Sound Design: an analy sis of the acoustics
and auditory effects of th e Data to Shape mapping
5. Digital Fabrication : the implementation of the Data to
Shape mapping in a CAD mesh which is 3D prin ted.
6. Acoustic Evaluation: comparison of the Acoustic
Sonification with the Baseline Prototype.
7. Perceptual Evaluation: a listening test to evaluate the
perception of known features in the dataset from the
Acoustic Sonification.
8. User-centred Evaluation: testing the usefulness of the
acoustic sonification in specific tasks.
7. REFERENCES
[1] G. Wolf. "Quantified Self".
http://www.webcitation.org/66TEHdz4d. Archived from
the original on 2012-03-26. Retrieved 2014-08-01.
[2] G. Wolf and K.K Kelly. Quantified Self Blog,
http://www.quantifiedself.org. Retrieved 2014-08-06.
[3] M. Carrigan. “Qualitative Self-tracking and the Qualified
Self”, http://markcarrigan.net/2014/07/23/qual itativ e-self-
tracking-and-the-qualified-self/. R etrieved 2014-08-06.
[4] J. Davis. “The Qualified Self”,
http://thesocietypages.org/cyborgology/2013/03/13/the-
qualified-self/. Retrieved 2014-08-06.
[5] D. Lupton. “Beyond the Quantified Self: the Reflexive
Monitoring Self”, http://simplysociology.wordpress.com/.
Retrieved 2014-08-06.
[6] E. Ramirez. “What We’re Reading”,
http://quantifiedself.com/2014/08/reading-22/. Retrieved
2014-08-06.
[7] Y. Jansen, P. Dragicevic, and J.D. Fekete. “Evaluating the
efficiency of physical visualiza tions”, In Proceedings of the
SIGCHI Conference on Human Factors in Computing
Systems (CHI '13). ACM, New York, NY, USA , 2593-
2602. DOI=10.1145/2470654.2481359
http://doi.acm.org/10.1145/2470654.2481359
[8] R.A. Khot, L. Hjorth, and F. Mueller. “Understanding
physical activity through 3D printed material artifacts”,
in Proceedings of the 32nd annual ACM conference on
Human factors in computing systems (CHI '14). AC M,
New York, NY, USA, 3835-3844.
DOI=10.1145/2556288.2557144
http://doi.acm.org/10.1145/2556288.2557144
[9] S. Barrass, "Digital Fabrication of Acoustic
Sonifications", Journal of the Audio Eng ineering Society,
vol. 60, no. 9, pp 709-715, September 2012.
[10] A. Jense. “Possibilities for 3D Printing Musical
Instruments”, Masters Dissertation in Industrial Design
Engineering, University of Twente.
[11] CCRMA. “3D Printing for Acoustics Workshop”,
https://ccrma.stanford.edu/workshops/pa3d2012/2013/.
Retrieved 2014-08-06.
SoniHED – Conferen ce on So nificatio n of Health and Environmental Data 12 September York, UK
[12] A. Hassaei, “3D Printed Record”,
http://www.amandaghassaei.com/3D_printed_record.html.
Retrieved 2014-08-06.
[13] Wolter, J. (2013) “Gramohorn 3D printed Acou stic
Speaker”, http://www.gramohorn.com/, accessed 3
February, 2014.
[14] Left Field Labs. “Music Drop”,
http://musicdrop.leftfieldlabs.com/. Retrieved 2014-08-06.
[15] J. Miller. “3D Printed Muffler”,
http://jmillerid.com/wordpress/2013/03/vacuum-pump-
muffler/. Retrieved 2014-08-06.
[16] H. Lipson. “Fully Functional Loudspeaker is 3D Printed”,
http://www.cornell.edu/video/fully-functional-loudspeaker-
is-3-d-printed. Retrieved 2014-08-06.
[17] C. Reas and B. Fry. Processing. http://processing.org/.
Retrieved 2014-08-06.
[18] NM. McLachlan, B. Keramati Nigjeh and A. Hasell. The
Design of Bells with Harmonic Overtones, Journal of the
Acoustical Society of America, 114 (1), p 505-511.
[19] Meshlab, http://meshlab.sourceforge.net/. Retrieved 2014-
08-06.
[20] netfabb. http://www.netfabb.com/. Retrieved 2014-08-06.
[21] J. Bertin, Semiology of graphics, Univ ersity of Wisconsin
Press, 1983.
