Data Sonification on Visualizing.JP

Quotidian Record

Sun, 04 Jan 2026 00:00:00 +0900

Quotidian Record is a work by artist and researcher Brian House that transforms personal GPS location data into music, visual form, and a physical object: an analog record. It reorganizes everyday movement into relationships among time, space, and sound.

The work goes beyond data visualization. It turns data into something that can be heard, touched, and physically manipulated.

Background

House created the work from location data collected from his own smartphone over an extended period. Ordinary movement through daily life became the basis for a rule-based translation into sound.

Why It Matters

Quotidian Record asks how personal data relates to bodily experience. Instead of treating GPS logs as points on a map, it transforms them into a material artifact that can be played like a record.

Design Lessons

Data can be materialized, not only visualized.
Personal data can be made experiential through sound and touch.
Translation rules are part of the artwork.
Everyday traces can become a structured composition.

Summary

Quotidian Record is an important example of data physicalization and sonification. It turns mundane location data into an object that links movement, memory, sound, and material form.

Distance Between Stations as Song Length, Income as Number of Instruments: Brian Foo's "Two Trains"

Fri, 06 Nov 2026 00:00:00 +0900

Brian Foo’s “Two Trains” is a piece of music in which the distance between stations on New York’s 2 subway line sets the length of each section, and the median income around each station sets the number, type, and force of the instruments that play. The author writes, “The number and type of instruments are selected based on that station’s surrounding median income.” The song is made of 48 sections connecting 49 stations, and runs about four and a half minutes in total. Most of the notes come from the E major scale, derived from the subway’s door chime (the author writes, “Most of the notes”).

The work is the first track of Foo’s music series “Data-Driven DJ,” and was released in January 2015. What follows draws on the author’s work page and FAQ, the code and sample data on GitHub, and three third-party articles. Where the numbers differ between sources, they are shown side by side. Other works that turn data into sound are also read in Chemo Singing Bowl and Quotidian Record.

2015: Making a Song That Rides the 2 Train

Foo is a programmer and visual artist based in New York. Data-Driven DJ is a series of ten tracks, and its home page describes it as musical experiments that combine “data, algorithms, and borrowed sounds.” The video was uploaded to Vimeo on January 31, 2015. The track is released under CC BY 4.0, and the code under the MIT License.

Foo gives his reasons for choosing the 2 line on the work page. It runs through three boroughs (Brooklyn, Manhattan, and the Bronx) and along Broadway. And then there is a graph of income.

“When I was looking at a graph of income inequality along the 2 Train, it looked like the perfect song composition with a build-up, climax, and falling action.”

The graph of income inequality along the 2 line looked to him like the structure of a song, with a build-up, a climax, and a falling action. This can be read as using a visualization as a preview before composing.

Foo also names time as a reason. Sonifying data requires the passage of time. So rather than seeing everything at once in a graph, you experience the data in sequence, as if riding the train.

There are two datasets. Income is the median household income around each station, from the U.S. Census Bureau’s American Community Survey 2011 Release. Station locations are the latitude and longitude from NYC Open Data’s “Subway Entrances.” The 49 stations run from Flatbush Ave.–Brooklyn College to Wakefield–241 St.

Distance Becomes Length, Income Becomes a “Budget”

According to the work page, the actual ride takes about 1 hour 45 minutes. The song compresses this to about four and a half minutes, which corresponds to a train running without stopping at about 336 miles per hour (540 km/h). In the code (subway.py), the number of beats in a section is the distance in meters divided by 75, rounded. The tempo is 120 BPM, so one beat is 0.5 seconds. That is 150 m per second, which matches 540 km/h.

The way income is mapped has the shape of shopping. According to the README, a station’s monthly income (the annual figure divided by 12) is that station’s “budget.” In the instrument table, each instrument has a “price” from 0 to 400. The script goes down the table from the top, buys an instrument if its price is lower than the remaining budget, and subtracts the price from the budget. When it hits an instrument that is too expensive, it stops there.

The other mechanism is the income percentile. For each instrument, the range of stations where it can be used is set by rank among the 49 stations. In the sample table, the same clarinet has a maximum volume of 0.2 for stations in percentiles 0 to 60, and 0.4 for stations in percentiles 60 to 100. Vocals 1 has three tiers, with the maximum volume rising through 0.05, 0.1, and 0.2.

The work page says that when the train passes through wealthy areas such as the financial district, the number of instruments, the volume, and the forcefulness increase.

A subway platform at Wall Street station. A station-name sign is on the left, and on the right a man leans out of a train window

From the work page of Data-Driven DJ “Two Trains” (photographer not credited)

The table below summarizes the mapping between data and sound, based on the work page and the code.

Data	How it is expressed in sound	Rule / value	Source
Distance to the next station	Length of the section	Beats = distance (m) ÷ 75. 120 BPM	Work page, subway.py
Income around the station (monthly)	Number of instruments that play	Monthly income is the budget; instruments are bought from the top of the table	subway.py, README
Income percentile	Type of instrument and volume cap	The higher the range, the higher the maximum volume	Sample table
Income (author’s summary)	Number and dynamics of instruments	3 to 30 instruments in each section	Work page
— (passage of time)	Fluctuation of volume and tempo	Sine curve with a period per instrument	subway.py
— (jitter)	Offset of note timing	±20 ms. Halton sequence	subway.py
— (pitch)	Scale and chords	Mostly E major notes	Work page
Borough	Not used in the sound	Shown in the accompanying video	subway.py

A scene from the video. On the left is a map with the red line of the 2 train drawn on it; on the right are the borough name “BRONX,” a red dot on a horizontal line, and the station name “Intervale Ave.”

Brian Foo (video still) / From Hyperallergic

The names of the boroughs are shown not in the sound but in the accompanying video, made with Processing.

Income Is Not Mapped Directly to Pitch: The Chime and Steve Reich

The scale of the song is chosen for reasons separate from income. According to the work page, the subway’s door chime comes from the two notes G# and E of the Westminster Chimes. Of the key built on them, E major (E, F#, G#, B), the author writes, “Most of the notes in my song also come from this scale.” In other words, most of the notes in the song come from this scale.

The stylistic model is Steve Reich’s “New York Counterpoint,” a piece for eleven clarinets. The 40 clarinet and bass clarinet samples in Two Trains play notes from D major and E major chords.

The work page lists phase shifting as one of the pillars of the song. It is a technique in which the same melody is repeated at slightly different tempos, drifting apart and coming back together, and the author explains that Terry Riley and Reich pioneered it in the 1960s.

“I felt this was the perfect metaphor for the NYC Subway: constantly looping but at different tempos, always running but never on time, phasing between order and chaos.”

Looping at different tempos, always running but never on time: this was the perfect metaphor for the subway. In the code, this fluctuation is made with sine curves. It is the style of the song, not data.

The composition is entirely algorithmic, and the work page says, “I did not manually compose any part of this song.” The sound material is 63 samples: 16 from songs by New York musicians, 40 clarinet-family samples, 6 xylophone samples, and 1 train horn.

So why did Foo choose volume and force to represent income? He writes:

“To be as objective as possible, the same rules are applied throughout the whole song. Also, I tried to select agnostic sound traits (e.g. volume, force) to correlate to median income rather than biased ones (e.g. sad vs happy sounds, vibrant vs dull sounds) to further let the data “speak for itself.””

In short: to be objective, the same rules are applied from the beginning of the song to the end. For income, he chose neutral sound traits such as volume and force, rather than biased ones such as sad versus happy. Sound traits that change mood were taken out of the representation of income.

As far as the code shows, what income changes is which samples play and the cap on volume. The sample file names include playing styles from pianissimo to fortissimo. The differences in force can be read as arising from both the samples that could be bought and the volume cap.

As for pitch, the code shows no rule that assigns income directly to pitch, as far as it can be read. However, the samples that play change with income, and each sample has its own pitch (Gs4, E3, D2, and so on). So the combination of pitches you hear changes with income indirectly.

On the other hand, the series FAQ describes an advantage of expressing data as music with the words “the creator can manipulate mood.” The series says it uses mood, while this song says it does not use mood for income. There can be read to be a tension between the two statements.

“3 to 30 Instruments” and “1:37”: Numbers That Differ by Source

The work page says that “between 3 and 30 instruments” play in each section. When this article’s researcher recounted the sample table on GitHub using the same rules as subway.py, a different number came out. It is not the author’s number, but a reference value.

Source	What was counted	Value
Work page (author)	Instruments playing in each section	3 to 30
Researcher’s recalculation	Rows of the instrument table assigned to one station	9 (E 180 St. and others) to 62 (Wall St., Chambers St., and others)

The two can be read as not counting the same thing. The table includes vocals with a price of 0, and the same timbre appears as separate rows for each percentile. The work page links to a Google spreadsheet as the final data, but I have not checked whether the sample table is the same as that.

The times differ as well. The author writes that the loudest part is Park Place to Chambers St. (income $205,192) at 1:37, and the quietest part is E 180 St. to Bronx Park East ($13,750) at 3:53.

Source	Subject	Value
Work page (author)	Length of the song	About 4.5 minutes
Vimeo video	Length of the video	286 seconds
Researcher’s recalculation	Length of the main body	261.5 seconds
Work page (author)	Loudest part	1:37
Researcher’s recalculation	Start of the Park Place→Chambers St. section	1:16
Work page (author)	Quietest part	3:53
Researcher’s recalculation	Start of the E 180 St. section	3:32

Both recalculated times are about 21 seconds earlier than the author’s. The video is about 24 seconds longer than the main body, which can be read as due to an introduction or margin at the start, but I have not confirmed the cause.

How Third Parties Saw the Song

The three articles look at different points. The first is a comparison with a precedent. Allison Meier of Hyperallergic (February 2015) set it beside Daniel Crawford’s “A Song of Our Warming Planet” (2013).

“Both projects make what can be abstract data into a narrative aural experience.”

Her assessment is that both turn data that can easily become abstract into a narrative listening experience. She also valued that, because the code is public, anyone can turn their own line into a song.

The second is a question from the music side. Molly Sheridan of New Music USA (July 2015) praised the choice of material, saying that anyone who knows Reich will hear the connection at once. She then asks:

“Should the music sound “better” when passing through rich neighborhoods in his presentation of income? … Where is the line between representation and judgment?”

Should the music sound “better” when it passes through rich neighborhoods, and where is the line between representation and judgment? Sheridan also reports that Foo does not retouch the song to make only parts of it sound good, and keeps the whole song faithful to the data.

The third concerns the relationship with visualization. Science Friday (February 2017) cited Foo as an example and wrote:

“so-called “data sonifiers” don’t expect their work to replace data visualizations, but rather complement the facts and figures they depict”

The view is that people who turn data into sound aim to complement visualizations, not to replace them.

Takeaways

There are four judgments to take back to visualization work.

The first is to narrow down the elements that carry data and fix the rest. In Two Trains, the length of time represents distance, and the number and force of instruments represent income. Phase shifting and the jitter of note timing are style, unrelated to the data. The scale, too, can be read as not directly assigned to income. It can be read as the same construction as a bar chart that uses only length for data and fixes color as decoration.

The second is to assign a variable that invites value judgments a trait that does not sway mood. Even so, Sheridan’s question remains. It can be read that it is not possible to fully avoid a loud sound seeming “rich.” In charts too, I think it is worth checking whether the brightness of a color looks like good versus bad.

The third is to place ordered data on a time axis. A line has an order of stations, and that order becomes the time of the song as it is. In expressions that use a time axis, it can be read as good to first check whether the data has a natural order.

The fourth is to publish the rules and the code, and to attach how things were counted and the standards used. Because the code and sample data were public, a recalculation was possible. As a result, it became clear that the author’s numbers and the recalculation from the sample data do not match. I have not confirmed the cause. What was counted, and whether the times include the introduction: if that much is written down, I think listeners can check the numbers against each other.

References

Primary sources

Brian Foo, Data-Driven DJ (series description, table of contents, Press section). https://datadrivendj.com/
Brian Foo, “Two Trains — Sonification of Income Inequality on the NYC Subway” (work page). https://datadrivendj.com/tracks/subway/
Brian Foo, Data-Driven DJ FAQ. https://datadrivendj.com/faq/
beefoo/music-lab-scripts (GitHub repository). https://github.com/beefoo/music-lab-scripts
README of 01_subway. https://github.com/beefoo/music-lab-scripts/tree/master/01_subway (raw: https://raw.githubusercontent.com/beefoo/music-lab-scripts/master/01_subway/README.md)
subway.py. https://raw.githubusercontent.com/beefoo/music-lab-scripts/master/01_subway/subway.py
data/stations.csv.sample. https://raw.githubusercontent.com/beefoo/music-lab-scripts/master/01_subway/data/stations.csv.sample
data/instruments.csv.sample. https://raw.githubusercontent.com/beefoo/music-lab-scripts/master/01_subway/data/instruments.csv.sample
LICENSE (MIT). https://raw.githubusercontent.com/beefoo/music-lab-scripts/master/LICENSE
Vimeo video (uploaded 2015-01-31, 286 seconds). https://vimeo.com/118358642
SoundCloud (audio). https://soundcloud.com/functionfoo/two-trains-sonification-of-income-inequality-on-the-nyc-subway
Data source: U.S. Census Bureau, American Community Survey 2011 Release. http://www.census.gov/acs/www/data_documentation/2011_release/
Data source: NYC Open Data, Subway Entrances. https://data.cityofnewyork.us/Transportation/Subway-Entrances/drex-xx56

Secondary sources

Allison Meier, “Data-Driven Music for the Disharmony of New York’s Income Inequality,” Hyperallergic, 2015-02-10. https://hyperallergic.com/180725/data-driven-music-to-for-the-disharmony-of-new-yorks-income-inequality/
Molly Sheridan, “Data-Driven DJ Brian Foo: Statistics That Sing,” New Music USA, 2015-07-22. https://newmusicusa.org/?p=277105
“How to Listen to Data,” Science Friday, 2017-02-07. https://www.sciencefriday.com/articles/how-to-listen-to-data/
“Hear the Sound of NYC’s Income Gap on This Data-Driven Subway Ride,” VICE (Creators Project). https://www.vice.com/en/article/hear-the-sound-of-nycs-income-gap-on-this-data-driven-subway-ride/

Turning a Year of Chemotherapy Blood Pressure into a Bowl That Rings: Stephen Barrass's Chemo Singing Bowl

Thu, 05 Nov 2026 00:00:00 +0900

Stephen Barrass’s Chemo Singing Bowl maps a year of blood pressure readings from a person undergoing chemotherapy onto the shape of a bowl. The data is conveyed through the sound the bowl makes when it is struck. The bowl is 3D-printed in stainless steel, and you sound it by tapping it or rubbing its rim. The artist calls a sounding object whose form is determined by data Acoustic Sonification.

The work was published on the artist’s blog on January 18, 2016, and in the same year was shown in an exhibition at the Museum of Applied Arts and Sciences (MAAS) in Sydney. What follows draws on the artist’s blog, the installation paper at ICAD 2016, a 2014 paper about the predecessor, and a 2019 paper on which Barrass is a co-author. Where the sources differ in how they describe things, I show the difference as it is.

2013: Turning his own blood pressure into a “singing bowl”

Chemo Singing Bowl has a predecessor: the Hypertension Singing Bowl, whose form was determined by the artist’s own blood pressure. It was published on the blog in July 2013 and written up as a paper at the SoniHED conference in 2014.

It began with a diagnosis of high blood pressure. Barrass started measuring his systolic and diastolic pressure with a cuff and logging it in an app. In the paper, he writes this about the time-series graph the app drew:

The plots of my blood pressure are generic and could have been share prices or CO2 emissions. Surely there was a way to make this personal dataset more personal, and more engaging.

In short: the graph of his blood pressure is generic, and could just as well show share prices or CO2 emissions; could this personal data be made more personal and more engaging?

For the sounding object he chose the Tibetan singing bowl, a bowl that “sings” with a sustained tone when its rim is rubbed. Singing bowls are associated with meditation and relaxation therapy, and these can lower blood pressure, which was his reason.

The first prototype (v0.0) varied wall thickness with the data, up to 5 mm. When struck it rang at 3628 Hz for 3 seconds, but it did not sing when rubbed. In the next version (v1.0) he made the wall a uniform 2 mm and placed the time axis around the circumference of the bowl. Each pair of readings was assigned one “spoke” joining the rim to the base, with the radius of the upper half varying with the systolic value and the radius of the lower half with the diastolic value.

The predecessor, Hypertension Singing Bowl. A dark metallic bowl with thin vertical bands lined up at regular intervals on its lower half

Source: Stephen Barrass, “Research Through Design into Acoustic Sonification”, ICAD 2016 (CC BY-NC 4.0)

v1.0 is 100 mm in diameter and weighs 162 g. Struck, it rings at 609 Hz for 10 seconds, and rubbed on the rim it sang like a traditional bowl. It also produced a 2 Hz tremolo and a hiss, which the artist says “may be caused by the data spokes.” A 2014 video of the predecessor being played is on YouTube. The ICAD 2016 paper sums up the predecessor this way:

The bowl sings like a traditional singing bowl, but with a timbre that is changed by the dataset.

In short: the bowl sings like a traditional bowl, but its timbre is changed by the data.

A year of chemotherapy, written into curved tines

The data in Chemo Singing Bowl is not the artist’s own. It is the blood pressure that a person treated with chemotherapy for breast cancer recorded over the year of treatment. The artist’s blog (2016) introduces it like this:

This version of the acoustic sonification singing bowl introduces 3D splines as a way of mapping the data into the form. The irregularity of the dataset makes for inharmonic and dissonant acoustic effects.

In short: this version introduces 3D splines (smooth curves drawn through control points) as the way to map data into form, and the irregularity of the data produces inharmonic, dissonant sound. The ICAD 2016 paper states the intent behind the mapping:

This mapping aims to amplify the effect of the dataset on the acoustics of the bowl by mapping the diastolic/systolic pressure to control points for the curvature of 2D splines amped to the tines.

In short: diastolic and systolic pressure become the control points that set the curvature of splines, and those splines are mapped onto the tines (the bands), in order to amplify the data’s effect on the acoustics (“amped” appears to be a typo for “mapped”). So the predecessor mapped data to thickness, while Chemo maps it to how the tines curve.

A photo of a gold bowl seen obliquely from above. Dozens of thin tines curve outward in arcs between the rim and the base

Source: Stephen Barrass / SONIFICATION.COM, “Chemo Singing Bowl”

In the photo, the tines are not uniform in length or in how they curve.

A photo of the bowl seen from directly above. Tines of uneven length and curvature stick out radially beyond the rim

Source: Stephen Barrass / SONIFICATION.COM, “Chemo Singing Bowl”

The correspondences that can be confirmed from the sources are as follows.

Data	Property of the object	Source
Blood pressure recorded over a year of chemotherapy	The shape of the whole bowl (3D-printed stainless steel)	Artist’s blog (2016)
Diastolic and systolic pressure values	Control points that set the curvature of splines; the splines are mapped onto the tines	ICAD 2016
A year of blood pressure readings	Variation of the splines around the bowl	Wang et al. (2019)
Irregularity of the data	Inharmonic, dissonant sound	Artist’s blog (2016)
How it is read	Sounded by striking or rubbing; the sound follows from the form	Wang et al. (2019)

No source states explicitly that the time sequence was laid out around the circumference. Since the predecessor v1.0 did so, and since the source speaks of “variation of the splines around the bowl,” Chemo can be read as also arranged around the circumference. The number of readings, the number of tines, the dimensions, and the frequencies were not found in any source.

A photo of the bowl held in a hand, seen from the side. The tines project outward from the rim and wrap around toward the base

Source: Stephen Barrass / SONIFICATION.COM, “Chemo Singing Bowl”

Does the bowl sing? Where the sources disagree

The sources differ on two points: whether the bowl sings, and whether the splines are 2D or 3D.

Point	Source	What it says
Does it sing?	ICAD 2016	“This bowl does not sing very well”
Does it sing?	Wang et al. (2019)	“not sing harmonically”
Does it sing?	Artist’s blog (2018)	“3D printed in stainless steep [steel] so it sings”
Dimension	Artist’s blog (2016)	“3D splines”
Dimension	ICAD 2016	“2D splines”

The 2018 blog post is a list of demos the artist brought to a seminar at Schloss Dagstuhl in Germany, and it uses the same wording for the predecessor.

The explanations for why it does not sing also differ. ICAD 2016 says “possibly due to the curved shape of the tines.” The blog and the 2019 paper point to the irregularity of the data. ICAD 2016 goes on to say that the next version would explore improving the singing effect using acoustic theory, but I found no source describing that version.

I found no source that says whether the shape of the tines or the irregularity of the data is the main cause. Because the data is mapped to curvature, the more irregular the data, the more irregular the tine shapes become, so the effect of the form and the effect of the data come out together in a single sound; this can be read that way. Set side by side, the predecessor mapped data to thickness and sang, while Chemo mapped it to curvature, for amplification, and sings less well; this suggests that both the character of the data and the choice of mapping show up in the sound.

“How I felt at the time”: the words of the person who supplied the data

According to the 2019 paper, the bowl was made as a gift for the data provider’s mother, in thanks for caring for her during the chemotherapy. The data provider’s reaction on hearing the sound is worded differently in three sources.

Source	Reaction
ICAD 2016	“it sounds as sick as I felt at the time”
Wang et al. (2019)	“that’s exactly how I felt at the time”
Artist’s blog (2016)	“… quite accurate – the sound is dry and heavy, without much resonance, just like I felt!”

In short: it sounds as sick as I felt then; that’s exactly how I felt then. The blog version continues: quite accurate, the sound is dry and heavy, without much resonance, just like I felt.

The predecessor was a bowl in which the artist listened to his own data. The SoniHED 2014 abstract says it invites a meditative reflection on the data and reminds one of a healthy life, and calls it a “poetic alternative” to the generic graphs of self-tracking.

Chemo was made from someone else’s data, and the owner of the data listened to it. The predecessor’s question of making personal data more personal came back, with another person’s data, to that very person; this can be read that way. The words the owner used were “accurate” and “exactly.” They can be read as pointing not to the correctness of reading a graph but to a match with a remembered bodily feeling.

How did third parties see this work?

I found no independent critique or press coverage of Chemo Singing Bowl. It is treated in three ways.

The first is the 2019 paper in IEEE Computer Graphics and Applications, “An Emotional Response to the Value of Visualization.” Co-authored by Yun Wang and others, with Barrass among them, it grew out of a working group at the 2018 Dagstuhl seminar.

Informed by the role of emotion in psychology, art, design, marketing, and HCI, we argue for an expanded definition of value, applicable to all forms of data visualization.

In short: drawing on the role of emotion, the authors argue for broadening the definition of visualization’s value. The paper writes that existing frameworks that measure efficiency (such as ICE-T) overlook the value of data physicalization, and it cites the Chemo bowl as an example. Because Barrass is a co-author, this is not a fully independent assessment.

The second is dataphys.org, a catalog of data physicalization works. It has no entry for Chemo alone, but has an entry for the predecessor group, “2011 – Stephen Barrass’ Physical Data Sonifications,” added by Pierre Dragicevic and recommended by Yvonne Jansen.

Physical representations of data can target other senses than vision or touch.

In short: physical representations of data can address senses other than vision or touch. About the HRTF Bells (bells whose shapes are determined by the characteristics of the left and right ears), the entry says the differences in form are barely visible, but the pitch and timbre can be compared.

The third is selection by a museum. MAAS curator Matthew Connell chose this bowl for “Out of Hand: Materialising the Digital,” which ran from September 3, 2016 to June 25, 2017. According to the artist’s blog, it was an exhibition on the role of digital fabrication, and the bowl was shown alongside works by Zaha Hadid, Iris van Herpen, and others. It was also presented at the DESIGN Canberra Festival that year, and was placed as a hands-on demo you could pick up and sound at ICAD 2016 and at Dagstuhl in 2018.

People picking up and sounding bells and a medal on a table. A dark bowl is on the right, and part of a bowl with gold tines can be seen at front right

Source: Stephen Barrass / SONIFICATION.COM, “Data Physicalisation @ Schloss Dagstuhl”

In book form, there is a chapter by Barrass, “Chemo Singing Bowl,” in Making with Data: Physical Design and Craft in a Data-Driven World, edited by Samuel Huron, Till Nagel, Lora Oehlberg, and Wesley Willett (AK Peters / CRC Press, 2022). It is chapter 3.1 in Part 3, “Digital Production,” pp. 185–198, DOI 10.1201/9781003264903-11. I have not been able to read the text of the chapter.

Takeaways

What this bowl lets us take away is that changing the mapping changes how the object itself behaves. With a graph on a screen, switching from showing a value as thickness to showing it as curvature does not change how the screen behaves. With a sounding object, the mapping directly determines the sound.

In the paper on the predecessor, as a reflection on v0.0 not singing, Barrass writes that it would have been more efficient to print a baseline bowl with no data in it first and test that, before moving on to the mapping. For physicalization where the material is the medium, this procedure can be used as is.

Chemo also shows that a mapping which “amplifies” the effect can undermine the foundation of a metaphor. The metaphor of a singing bowl depends on the bowl singing. Even so, the owner of the data called the muddy sound “accurate.” The collapse of the metaphor itself can be read as having conveyed the data.

The other point is separating causes. When you change the mapping, also make a version with no data in the same form and compare by ear. Then you can say how much of the audible difference is due to the data. That is what I think.

References

Primary sources

Stephen Barrass, “Chemo Singing Bowl” (SONIFICATION.COM, 2016-01-18) https://sonification.com/2016/01/18/chemo-singing-bowl/
Stephen Barrass, “Research Through Design into Acoustic Sonification”, ICAD 2016 installation paper (CC BY-NC 4.0) https://www.icad.org/icad2016/proceedings2/installations/ICAD2016_paper_52.pdf
University of Canberra Research Portal, “Chemo Singing Bowl: Out of Hand - Materialising the Digital” https://researchprofiles.canberra.edu.au/en/publications/chemo-singing-bowl-out-of-hand-materialising-the-digital/
Stephen Barrass, “3D Printed Acoustics @ Out Of Hand, Sydney” (2016-09-03) https://sonification.com/2016/09/03/out-of-hand-materialising-the-digital/
Stephen Barrass, “Chemo Singing Bowl @ DESIGN Canberra” (2016-10-28) https://sonification.com/2016/10/28/design-canberra/
Stephen Barrass, “Data Physicalisation @ Schloss Dagstuhl” (2018-10-30) https://sonification.com/2018/10/30/data-physicalisation-schloss-dagstuhl/
Stephen Barrass, “Making With Data” (2022-12-13, reprint of the table of contents) https://sonification.com/2022/12/13/making-with-data/
Stephen Barrass, “Chemo Singing Bowl”, in S. Huron, T. Nagel, L. Oehlberg, W. Willett (eds.), Making with Data: Physical Design and Craft in a Data-Driven World, AK Peters / CRC Press, 2022, pp. 185–198 (text not read) https://doi.org/10.1201/9781003264903-11
Stephen Barrass, “Acoustic Sonification of Blood Pressure in the Form of a Singing Bowl”, SoniHED 2014, York https://www.york.ac.uk/media/c2d2/media/sonihedconference/Barrass_SoniHED_2014.pdf
University of Canberra Research Portal, “Acoustic Sonification of Blood Pressure in the Form of a Singing Bowl” (abstract) https://researchprofiles.canberra.edu.au/en/publications/acoustic-sonification-of-blood-pressure-in-the-form-of-a-singing-/
Stephen Barrass, “Hypertension Singing Bowl” (2013-07-10) https://sonification.com/2013/07/10/hypertension/
Stephen Barrass, demo video of the Hypertension Singing Bowl (YouTube, 2014) https://www.youtube.com/watch?v=DE3tolD5Krg
Stephen Barrass, “Sound in Data Physicalisation”, Dagstuhl Reports 8(10), 2019, §5.4 https://drops.dagstuhl.de/storage/04dagstuhl-reports/volume08/issue10/18441/DagRep.8.10.127/DagRep.8.10.127.pdf

Secondary sources

Yun Wang, Adrien Segal, Roberta Klatzky, Daniel F. Keefe, Petra Isenberg, Jörn Hurtienne, Eva Hornecker, Tim Dwyer, Stephen Barrass, Theresa-Marie Rhyne, “An Emotional Response to the Value of Visualization”, IEEE Computer Graphics and Applications 39(5), 2019 (Barrass is a co-author) https://petra.isenberg.cc/publications/papers/Wang_2019_AER.pdf / doi:10.1109/MCG.2019.2923483
dataphys.org, “2011 – Stephen Barrass’ Physical Data Sonifications” (added by Pierre Dragicevic, recommended by Yvonne Jansen) https://dataphys.org/list/barrass-physical-data-sonifications/
Jason Alexander, Petra Isenberg, Yvonne Jansen, Bernice E. Rogowitz, Andrew Vande Moere (eds.), Dagstuhl Seminar 18441 “Data Physicalization” report, Dagstuhl Reports 8(10), 2019 https://drops.dagstuhl.de/storage/04dagstuhl-reports/volume08/issue10/18441/DagRep.8.10.127/DagRep.8.10.127.pdf
Out of Hand: Materialising the Digital (MAAS, curator Matthew Connell, 2016-09-03 to 2017-06-25). The selection is confirmed on the artist’s blog https://sonification.com/2016/09/03/out-of-hand-materialising-the-digital/
Making with Data book site https://makingwithdata.org/

Assign One Column to Rungs of a Musical Scale, Play One Row per Beat: TwoTone

Fri, 06 Nov 2026 00:00:00 +0900

TwoTone is a tool that assigns one column of a table to rungs of a musical scale, from the minimum to the maximum value, and plays each row as one beat. The data decides only the pitch and whether a note sounds or stays silent. By default, values are rounded onto 15 rungs, two octaves of C major. Volume, instrument and key are chosen by the user for each track. For tempo, the user sets the overall BPM and a per-track multiplier. It is a free web app that runs in the browser and needs no code.

TwoTone was released on March 5, 2019. It was made by Datavized Technologies, a data design studio in New York, with support from the Google News Initiative (GNI). Sonify took over maintenance in December 2021, and development stopped in mid-2023. What follows draws on the GitHub repository (README and source code), the official site twotone.io, and articles published by people involved in its development.

March 2019: Released for the newsroom

TwoTone was commissioned by Simon Rogers, then data editor at Google News Lab. Alberto Cairo of the University of Miami served as adviser, and development was led by Datavized CEO Hugh McGrory. On the day of release, Rogers wrote:

“Previously, to make data into music, you’d need to have expert coding skills. Now, anybody can do it.”

In other words, turning data into music used to require advanced programming, and now anyone can do it.

The sources describe the developer differently.

Source	How the developer is described
McGrory, “An Intro to TwoTone” (2019)	“imagined by Datavized with support from Google News Initiative”
Official site (twotone.io/about)	“originally made by Datavized Technologies with support from Google News Initiative and is now maintained by Sonify”
Current README (sonifydata/twotone)	“a Sonify project built with support from Google News Initiative”, followed by “originally developed by Datavized Technologies”

What followed is as below.

Date	Event	Source
2019-03-05	Release. McGrory, Rogers and Cairo each publish an introductory article	Each article
2021-12-10	The repository moves from datavized/twotone to sonifydata/twotone	README of datavized/twotone
2022-01-14	MIDI output added	git log
2023-10-12	“As of mid-2023, development on TwoTone has been paused.” added to the top of the README	README, git log
2024-07-27	Last commit (adds sample data)	git log

Sonify is a US company co-founded by McGrory. As of November 6, 2026, the old app at datavized.github.io/twotone/ opened. sonify.io/twotone/, the address the README points to, did not resolve.

McGrory writes that the idea came from the 2016 TED talk by Wanda Diaz Merced, a blind astronomer. Granular synthesis worked, but it was dropped as “too complex for the beginner”.

Rounding one column onto 15 rungs of a scale

The user first loads a table. The accepted formats are .xls, .xlsx, .csv and .ods, with limits of 20MB and 2,000 rows. Data is processed only inside the browser and is not sent to a server.

The “Select Data Source” screen. Seven sample datasets, including how people in the Basque Country spend their time and US honey production, are listed along with the 20MB and 2,000-row limits

Source: Datavized Technologies / Hugh McGrory, “An Intro to TwoTone” (Medium), one frame of a GIF animation. No license is stated for the image (MPL 2.0 is the license of the code, and the text does not say it applies to images).

Once data is loaded, TwoTone automatically creates one “Musical Scale” track. Checking the source code, a value becomes a note in two steps. First, the value is stretched to the range 0 to 1 using the column’s minimum and maximum. Then it is multiplied by “number of rungs minus 1” and rounded, which gives the rung of the scale. The default number of rungs is 15, and a code comment calls this “two whole octaves (inclusive)”. The README describes the two ends as follows.

“The lowest and highest values will play the tonic note at either end of the given scale.”

That is, the minimum and maximum play the tonic note at either end of the scale. With the default C major, the minimum is the lower C and the maximum is the C two octaves higher.

Data	Sound	How it is decided	Source
Row order	Time	One row is one beat. The length of a beat is set by the overall tempo (BPM)	README “Track Tempo”
Values of the chosen column (numbers, dates)	Pitch	Min to max is mapped to 0 to 1, then rounded to a rung with `round(value × (number of rungs − 1))`	data.worker.js, scale/index.js
Text column	Pitch	The distinct values in the column are lined up, and their rank is mapped to 0 to 1	data.worker.js
Blank cells, non-numeric values	Rest	No sound is played	scale/index.js
Range of another column (filter)	Sounds / silent	Rows outside the range are silenced	README “Filtering a Track”

The TwoTone editor. Two tracks appear as bar charts on the left and a table by year on the right. The row for 2005, which is playing, is shaded gray, and the corresponding bar is orange

Source: Datavized Technologies / Hugh McGrory, “An Intro to TwoTone” (Medium). No license is stated for the image (MPL 2.0 is the license of the code, and the text does not say it applies to images).

The upper track on screen plays the “Value of Production ($)” column of the honey production data. The bar chart shows the value of each row, and the bar of the row being played is orange. As an aside, the README’s example says that 100 rows at 60 bpm take “100 seconds, or 1:20”, but 100 seconds is 1 minute 40 seconds.

Rungs are counted as steps of the scale, not as semitones. For that reason, small differences between values can be heard only down to the number of rungs, as I read it. The README says:

“More octaves will result in finer-grained distinction between values. Fewer octaves will allow different tracks to represent different data at opposing ends of the scale, even using the same instrument, like playing left hand and right hand on a piano.”

More octaves separate values more finely, and fewer octaves let different tracks occupy the upper and lower parts of the range.

Another column decides sound or silence

Besides pitch, the data decides whether a note sounds. Every track has a filter; when the user picks a column and a range of values, rows outside the range become silent. The bar chart shows which parts will sound.

The expanded view of a Musical Scale track. The left side of the bar chart is gray, and below it are settings such as Volume, Filter Value, Key (C, Major) and Scale Range (2 Octaves)

On screen, the Filter Value handle has moved to the right, and the left side of the bar chart is gray. Gray rows do not sound. On a “Narration Audio” track, which plays recorded speech, choosing Active sections plays the audio once at the start of each sounding section. This can be read as the data deciding when the audio is triggered.

There is also a feature that reads the title aloud with Google’s Text-to-Speech API and places it before the music. McGrory adds a caveat about this feature.

“Being pedantic, this means that Twotone can’t really be described as ‘sonification’ since that word refers to ’the use of non-speech audio to convey information or perceptualize data’.”

Since it uses spoken words, strictly speaking it cannot be called “sonification”.

Volume, instrument and key are chosen by people

The remaining track settings are chosen by people, not by the data. In the current source, the instruments are ten, including Piano, plus Midi Out. There are 12 keys and 11 modes, such as major, minor and dorian. The user also chooses the number of octaves, the starting octave, the tempo multiplier, the arpeggio (up, down, none) and the volume of each track.

Whether volume is tied to data is described differently in different sources.

Source	What it says about volume
README “What is TwoTone’s goal?”	“attaching … individual columns of data to variables in sound (pitch, volume)”
README “Volume Adjustment”	“Every track has a volume slider in the expanded view.”
src/engine/AudioDataEngine.js	Passes the stored `track.volume` directly as the volume (gain)
src/engine/types/scale/index.js	What is decided from a row’s value is only the pitch

I checked the source on master as of November 6, 2026.

The makers divide its uses into two. One is understanding data.

“If understanding the data is your goal, you may want to keep things simple and resist the urge to make too many layers or add too many effects.”

If understanding is the goal, the advice is not to pile on layers or effects. The other use is making music, and McGrory writes that musicians like it because “it adds a randomness that sparks new ideas”. This can be read as one tool housing both understanding data and making songs from chance.

Several examples appeared on the day of release. Cairo sonified US unemployment (2009 to 2018). Maarten Lambrechts played global temperature anomalies from 1880 to 2018 on a glockenspiel in C minor. Rogers invited readers to guess, from sound alone, when people in the Basque Country head out to work in the time-use data.

How third parties saw the tool

Rogers and Cairo were the commissioner and the adviser, so here I look at those who were not involved in its development.

The first is its assessment as an entry point for journalism. In 2024, the RJI (Reynolds Journalism Institute), in an overview of sonification tools for journalism, listed TwoTone first.

“Both playful and practical, it is an ideal way to explore turning data into audio.”

It is playful and practical, and suited as an entry point to turning data into sound. The article says that as the price of honey rises, a rising arpeggio responds accordingly.

The second is use in teaching. A lesson from the Sherman Centre for Digital Scholarship at McMaster University gives steps for using TwoTone on data-breach data. It says to silence every track except the one to be emphasized. This can be read as an example showing that emphasis through volume is a human operation, not linked to the data.

The third is awards. In 2019 it was a finalist in the technical innovation category of the Online Journalism Awards and was longlisted at the Information is Beautiful Awards. I could not confirm the judges’ comments for either. I did not find any academic evaluation devoted to TwoTone alone.

Takeaways

There are three judgments you can take back to visualization work.

The first is to treat the number of rungs as the resolution. With the default 15 rungs, values become only 15 distinct pitches. To let differences be heard, increase the octaves; when layering columns, separate their ranges. This can be read as the same judgment as choosing how fine a scale’s ticks are.

The second is the effect of stretching by minimum and maximum. Because the two ends become the tonic notes at either end of the scale, a single extreme outlier would, as I read it, crowd the remaining values into a few narrow rungs. It would be wise to look at the range of values before turning them into sound.

The third is to tell apart what the data decides and what people choose. In TwoTone, volume, instrument and key are not data. When presenting the sound, it is probably better to say up front, in the title or the description, what is data and what is staging.

From the same series, you can also read Chemo Singing Bowl, which turns blood pressure into the shape of a ringing bowl, and Quotidian Record, which turns a record of movement into a vinyl record.

References

Primary sources

sonifydata/twotone, README (retrieved 2026-11-06). GitHub / raw
Source code of sonifydata/twotone (src/constants.js, src/util/data/data.worker.js, src/engine/types/scale/index.js, src/engine/types/scale/samplerInstruments.js, src/soundq/src/util/scales.js, src/engine/AudioDataEngine.js, src/engine/speech.js, LICENSE, git log). GitHub
datavized/twotone (notice of the move). GitHub
TwoTone official site. Top / About / How it works / Examples / Tutorials
Hugh McGrory, “An Intro to TwoTone,” 2019-03-05. Medium
Simon Rogers, “Sonification: make beautiful music with your data,” 2019-03-05. Medium
Alberto Cairo, “Sonifying data with TwoTone,” The Functional Art, March 2019. Blog
Tutorial videos (embedded in Cairo’s blog). YouTube 1 / YouTube 2
Sonify. Top / TwoTone
Online Journalism Awards 2019, TwoTone entry page. Page
Information is Beautiful Awards 2019, TwoTone showcase page. Page
TwoTone app. Old app / MIDI output beta

Secondary sources

Lamar, “Embracing data-driven audio as a storytelling method,” RJI (Reynolds Journalism Institute), 2024-08-08. Article
Sherman Centre for Digital Scholarship, McMaster University, “Data Sonification,” Lesson 4 (CC BY 4.0). Lesson

Making a Line to See and a Line to Hear from One Table: Highcharts Sonification Studio

Fri, 06 Nov 2026 00:00:00 +0900

Highcharts Sonification Studio (HSS) is a tool that turns the data you enter in a table into a line chart and, at the same time, a sound that traces that line. In a sidebar, without writing any code, you choose whether values map to pitch, panning, volume, duration, and so on. During development, three evaluations were run that involved totally blind users and experts with visual impairments.

It was developed by Highsoft AS (Norway), the maker of Highcharts, and designed jointly with the Georgia Tech Sonification Lab. It runs in a browser, is free, and requires no login. What follows is based on the ICAD 2021 paper (Cantrell, Walker, Moseng), the GitHub source and in-app tutorial (v1.1.1), and a Georgia Tech news article (July 2021).

2019: Moving the Java Sandbox to the browser

HSS has a predecessor: the Sonification Sandbox, a Java application that the Georgia Tech Sonification Lab maintained for about 15 years. It turned data worked on in a table into a chart you could both see and hear. The paper says that moving to the browser makes the tool easier to distribute and easier to use with screen readers. It also says that elaborate sound is harder to produce (“advanced audio can still be a challenge”). The paper calls HSS “a systematic re-implementation of the Sonification Sandbox,” and also notes that not every Sandbox feature was carried over.

The first GitHub commit is dated August 26, 2019, by Øystein Moseng of Highsoft. The paper’s three authors are Stanley J. Cantrell, Bruce N. Walker (director of the Sonification Lab), and Moseng. It was presented at ICAD in June 2021, and at the time of the paper (March 2021) the tool was a “public beta.” v1.0.0 came in September 2022, and the latest, v1.1.1, in November 2025.

The interface has two views, a table tab and a chart tab, with settings in a sidebar on the right.

The HSS screen: a play button on the left, a two-series area chart in the center, and global audio settings on the right

Source: Highsoft AS, sonification-studio (GitHub) (the footer in the screenshot reads v0.1.5-beta)

Mapping values to pitch, panning, volume, and duration

The landing page states its goal as follows.

“We want users to be able to use the tool without having to write code, and without prior sonification expertise.”

In other words, the aim is that neither coding nor knowledge of sonification should be required. You put data into the table by importing a CSV or typing it in, and choose a chart type such as line, area, scatter, or bar.

The top half shows a two-column table of X and Y; the bottom half shows a line chart connecting those values

Source: Highsoft AS, HSS tutorial (GitHub)

Sound settings are per series. For each parameter you choose from Default, Fixed (a fixed value), or Mapped (tied to the data). With Mapped, you set a minimum and maximum, the column to map from, and the direction (Mapping direction). Judging from the source, the only columns that can serve as the source of a mapping are x and y.

The Volume settings. Note volume is Mapped, minimum 0, maximum 100, Map volume to is y, Mapping direction is Low to high

Source: Highsoft AS, HSS tutorial (GitHub)

The correspondence between data and sound that can be read from the source and the tutorial is as follows.

Data	Sound parameter	Defaults, ranges, etc.
x or y	Pitch (Note pitch)	Minimum 36, maximum 90 (“number of notes from C0”). The direction can be reversed
x or y (default is x)	Panning (Note panning)	Controlled by “Default panning enabled” in the global settings, which is on by default. Can be overridden per series
x or y, or a fixed value	Volume	Overall volume defaults to 70%
x or y, or a fixed value	Note duration	Only has an effect on instruments that sustain, such as flute and trumpet
x or y	Low-pass / high-pass filter, tremolo depth and speed	―
Series	Instrument	Not mapped to data. Defaults to piano
Series	Sound on/off	You can remove the sound while keeping the visual

However, I could not confirm whether pitch corresponds to x or to y when set to Default.

The Highcharts library documentation, which sits underneath, lists tips for mapping as “7 tips from science.” One of them is “G4 to F6 is the most accurate pitch range for most humans.” The same document recommends “Dual coding,” layering the same value across multiple sounds, and cautions against relying on panning alone.

If C0 is counted as 0, HSS’s default of 36 to 90 corresponds to C3 through F#7, which can be read as a much wider range than the G4 to F6 the documentation recommends. This can be read as a default that favors a range wide enough for differences to be audible over accuracy.

Playback speed, and thinning that keeps the peaks

The global settings are gathered under “Global settings.”

Setting	What it does	Default, etc.
Play speed	Sets the total playback time	Default 79, about 8 seconds. Calculating from the formula, 100 gives about 2 seconds
Play detail	Sets the number of notes played. Changes neither the data nor the playback time	1 to 15, default 15
Global audio context cues	A tick sound in the background, based on X values	Instrument: chop
Play marker	Shows the current playback position as a crosshair on the chart	On by default
Text description	Reads out a sentence giving, for example, the number of points and the minimum and maximum of Y	―

The Basic section of the Audio global settings. Play Speed (duration 8s) is 79 and Play Detail is 15

Source: Highsoft AS, HSS tutorial (GitHub)

On Play detail, the tutorial says the following.

“When detail is low, and few notes are played, the tool will group data together and play each group as one. The minimum and maximum of each group will be played, to preserve peaks in the data.”

When the detail is lowered, points are grouped into intervals, and the minimum and maximum of each interval are played so that the peaks survive. The source likewise specifies a method (minmax) that picks the minimum and maximum of y within each interval.

According to the tutorial, the context cue plays a click at each X-axis tick so that the passage of time is easier to grasp. The context cue, the crosshair, and the text description can be read as devices for conveying position within the chart, whether by sound alone or by sight alone.

What was counted in three evaluations

The second half of the paper reports on a three-phase evaluation. The results of each phase were passed to Highsoft and used in designing the next version.

Phase	When, version	Who evaluated
Phase 1	March 2020, v0.0.4	A 33-task study with 12 sighted participants and 1 participant who identified as totally blind (a VoiceOver and JAWS user). Separately, 1 expert who identified as legally blind
Phase 2	November 2020, v0.0.9	4 experts (3 of whom identified as legally blind)
Phase 3	February 2021, v0.0.11	2 experts (1 of whom identified as legally blind). An informal evaluation in which Highsoft’s head of accessibility also joined, going through the screens in order

The 33 tasks in Phase 1 range from working with the table, to playing and stopping, adjusting volume and speed, and changing the chart’s title and colors. The researchers scored each task on a usability measure called PURE and reported the results as counts of tasks.

Participants	Tasks receiving the best rating (perfect PURE rating)	Remaining tasks
Sighted participants (12)	29 of 33 tasks (87.9%)	4 tasks rated lower
Totally blind participant (1)	13 of 33 tasks (39.4%)	20 tasks rated lower, of which 4 could not be completed without the researcher’s help

Both 87.9% and 39.4% are “the share of the 33 tasks that received the best rating.” They are not shares of people. The data from the totally blind participant was analyzed separately from that of the sighted participants. The four tasks that needed help were moving to the last cell of the table, entering data, stopping during playback, and pressing the button that reads out the description. The expert evaluations concluded that compatibility with JAWS and NVDA needed improvement.

Some of the experts’ comments concern wording.

“Be cautious about using the terms ‘frequency’ and ‘panning’ in the UI; these concepts cannot be assumed to be universally understood, especially students”

The point is that “frequency” and “panning” should not be assumed to be understood, especially by students. A Phase 1 task was “Adjust the minimum frequency to 10%.” In the current version the heading is “Minimum note,” and the tutorial likens pitch to piano keys. However, the help text still contains “pitch/frequency,” and I found no document explaining why the wording was changed.

How third parties viewed the tool

The first is the voice of a blind user. The Georgia Tech news article quotes Brandon Biggs, a blind researcher and entrepreneur.

“I love how accessible all the components are with a screen-reader and how easy it is to create a sonification,”

The sentiment is that every component works with a screen reader and that making a sonification is easy. But this is a remark inside an article published by a co-developer, and it is not an independent evaluation.

The second is the limit of its expressiveness. In a CHI 2024 paper, Hyeok Kim, Yea-Seul Kim, and Jessica Hullman described the Sandbox’s successors, HSS among them, as follows.

“Sonification Sandbox [53] and its successors [14, 27] support more encoding channels, but developers need to use external sound editors to sequence or overlay multiple sonifications”

The point is that many sound parameters can be assigned, but sequencing or overlaying the sounds you create requires outside software. This can be read as the flip side of a design that is completed through sidebar operations.

The third is adoption and sustainability. Sofia Vallejo Budziszewski and Katharina Gross-Vogt of IEM at the University of Music and Performing Arts Graz surveyed sonification software at ICAD 2025.

“Tools such as Tone.js [26], TwoTone [12], and Highcharts Sonification [27] have gained traction due to their ease of use and web-based deployment.”

They list it alongside TwoTone as an example that spread through ease of use and running in the browser. The same section also says that many tools from universities end for lack of funding and institutional backing. The HSS paper positioned itself as a free resource backed by a company that sells data tools. A company taking over maintenance can be read as one reason updates continue. The ICAD 2025 paper does not say so.

There are also points where the sources disagree.

Point	What each source says
License	The Georgia Tech news article (2021) says “MIT License.” A commit on September 7, 2022 rewrote the notice to BSD 3-Clause
Highcharts	The article says “open-source web application Highcharts.” LICENSE.md treats the Highcharts library as under a commercial license
Source release	The paper (as of March 2021) uses the future tense, “will be made opensource.” The first commit is from August 2019, and I could not confirm when it was made public

Takeaways

There are four decisions that can be taken back to visualization work.

The first is to build the visual and the sound from the same table, and to let either one be removed on its own. Even if a series’ sound is removed, the line stays. This can be read as a design in which people who look and people who listen share the same chart.

The second is to reduce the mapping to three choices: which column, over what range, in which direction. This can be read as carrying the domain, range, and direction of a visual scale straight over to sound. Whether to use a wide default pitch range or narrow it to the range the documentation recommends remains a choice left to the maker.

The third is to keep peaks and valleys when thinning points. Grouping by average smooths out extreme values. It is worth deciding in advance what the thinning is meant to preserve.

The fourth is to look at evaluation numbers split by whose they are and what they are a share of. On the same 33 tasks, the best rating went to 87.9% of tasks for sighted participants and 39.4% for the totally blind participant. With sighted participants alone, the result was that the tool was nearly easy to use. The paper’s numbers can be read as showing that some problems become visible only when blind people use the tool.

References

Primary sources

Highcharts Sonification Studio (official site). Site
highcharts/sonification-studio (GitHub. README, LICENSE.md, src/app/home/Home.vue, src/app/home/TutorialConfigureAudio.vue, the defaults and mapping source under src/app/editor/, and the git history, consulted at v1.1.1). Repository
Stanley J. Cantrell, Bruce N. Walker, Øystein Moseng, “Highcharts Sonification Studio: An Online, Open-Source, Extensible, and Accessible Data Sonification Tool,” ICAD 2021, pp. 211–216 (CC BY-NC 4.0). doi:10.21785/icad2021.005 / PDF / Georgia Tech repository
Joshua Preston, “New Browser-Based Chart Builder Gives Line Graphs, Scatterplots Their Own Audio Track,” Georgia Tech College of Computing, 2021-07-27. Article
Highcharts Docs, Sonification: Getting started / Mapping. Getting started / Mapping
HSS Tutorials (community site on the Georgia Tech side. Live demos by Bruce Walker, etc.). Site / YouTube
Sonification Sandbox (Georgia Tech Sonification Lab). Page

Secondary sources

Brandon Biggs’s remark (in the Georgia Tech news article above; an article from the co-developer). Article
Hyeok Kim, Yea-Seul Kim, Jessica Hullman, “Erie: A Declarative Grammar for Data Sonification,” CHI 2024. arXiv:2402.00156
Sofia Vallejo Budziszewski, Katharina Gross-Vogt, “Legacy, Active & Future Software Tools in Sonification Research,” ICAD 2025. PDF