What Just Happened? Yuri Suzuki Broadcasts the “Sound of the Earth” at Triennale Milano and in Your Headphones

Yuri Suzuki is an artist, designer and electronic musician. He explores sound and visual language through artworks and projects that examine the relationship between people and their environments, questioning how both visual art and sound evolve to create personal experiences. Suzuki has exhibited in solo and group exhibitions nationally and internationally, including at Tate Britain, the V&A, the Barbican, and MoMA. He joined Pentagram as a partner in 2018.


Greg J. Smith

What just happened? On July 15th at Triennale Milano, Yuri Suzuki unveiled Sound of the Earth: Chapter 3, a massive geodesic sphere (comprised of 300 speakers) that plays a megamix of crowdsourced geolocated field recordings. The project uses machine learning to mix similar sounds into a lulling never-ending ambient soundscape that you can enjoy on your headphones (via the project website), or on-site in Milan.

Q: You released your spherical record The Sound of the Earth (2012), which ‘placed’ field recordings by geography some years ago. What inspired that project, initially, and how has your thinking about geography and sound evolved since then, and informed the subsequent chapters of the project?
A: I became interested in collecting everyday field recordings when I was travelling around the world many years ago. This experience inspired me to want to share these sounds with others and transport people sonically, so I started Sound of the Earth to explore this further.

The COVID-19 lockdowns obviously really restricted how people were able to travel and this led me to think about how to share these sounds online. When we launched the online Pandemic Chapter with the Dallas Museum of Art it really expanded the project for me. I love working collaboratively, and hearing the crowdsourced sounds inspired me creatively—Chapter 3 takes this idea further by combining a digital site with the physical installation in Milan. Creating a hybrid online platform with a physical artwork is a really interesting experiment as people can choose how they experience and participate in the project.
“I became interested in collecting everyday field recordings when I was travelling around the world many years ago. This experience inspired me to want to share these sounds with others and transport people sonically.”
Q: The web interface is quite elegant and the ambient mix it serves up is fittingly dreamy. Could you talk a little bit about how machine learning is being used behind the scenes, and what design (composition) decisions are made to select and sequence recordings?
A: My frequent collaborators Counterpoint along with funding from Google’s Artists + Machine Intelligence grant allowed me to explore how machine learning can create ever changing soundscapes. Our submission app loads a TensorFlow.js version of the YAMNet model from TensorFlow Hub. It then analyzes each user submission on their own device before it’s uploaded to our system. Our database stores the sound file, as well as the YAMNet embeddings vector and classification. So the machine learning model is applied immediately upon each user submission and Counterpoint developed bespoke adjustments that happen afterwards.

My team and I composed an ambient backing track to help set the tone for the soundscape, we wanted something mellow which would allow the crowdsourced sounds to be highlighted as clearly as possible.
An auditory portrait of the world

Drawing on collaborations with Counterpoint (machine learning), Ableton (sound spatialization), and Xilografia (fabrication), and support from Google Artists + Machine Intelligence, Yuri Suzuki’s Sound of the Earth: Chapter 3 is an “auditory portrait of the world.” Geotagged field recordings are uploaded to a project website, their sonic characteristics are analyzed and categorized by machine learning, and the resulting archive of sounds are mixed together, drifting from sound-to-sound based on sonic likeness. The resulting ambient composition can be enjoyed via the geodesic 300-speaker monolithic sculptural form installed at Triennale Milano and online.

Q: I’m struck by the figure of the ‘global speaker’ in the Triennale installation. Does it mix sound by mapping geocoordinates to speaker coordinates on the sphere? Secondly, what technical challenges did you encounter when constructing this mega speaker?
A: I really wanted this chapter of the series to blur the boundaries of the geographical borders even more than the previous editions of the project which is why I chose to explicitly represent the geography via the submissions collected on the website. It’s great to see how many people contributed, and where they are all from, but I wanted the installation to focus on the sonic similarities rather than the geographical differences.

We are using approximate nearest neighbour searching to stumble around in the YAMNet embedding space. And then we have our own code for the soundscape generation that uses all this information to cue and play the sounds and add subtle effects.

Technically, we were assisted by the fantastic team at Ableton, who helped us translate the website sounds into the three-dimensional space of the Triennale installation. Using Ableton Live, we have sophisticated control of the visitors’ sonic experience. I really want people to physically follow the sonic pathways the machine learning creates when they visit the exhibition. The geodesic sphere was made in Milan by our fabricators Xilografia who have incredible precision, which is essential when creating such a complex form.
“I’d like to continue trying to bond people together with sound, as, sadly, it feels like the world is becoming increasingly disconnected.”
Q: You’ve been iterating this idea of sound and geography for years, what’s next for Sound of the Earth? Do you see any logical next steps for this project, either in its hardware or software forms?
A: The whole Sound of the Earth series is quite a personal project for me. Up until the age of 26 I didn’t have much opportunity to travel around the world, and it was such a great experience when I did finally got to visit and live in totally different cultures and climates. Of course my fascination is always with sound—I started to notice differences in the soundscapes as well as the music. I began to hear local sonic representations for the first time.

The Sound of the Earth project began as the result of these trips. I became interested and fascinated by the idea of a crowdsourced archive and how to create a platform to make it public. When I did the project for the Dallas Museum Art, I created a strong link with the website and the online world that became so important to us all during the pandemic. With Chapter 3 I have created this hybrid artwork which combines the physical installation working alongside the digital site. I’m not sure what the next step in the series will be, but I’d like to continue trying to bond people together, as, sadly, the world currently feels like it is becoming increasingly disconnected.
What Just Happened?

In this serial interview format, HOLO checks in with artists, designers, curators, and researchers to get the lowdown on a timely topic—be it a new project, exhibition, or current event that ‘just happened.’

What else ‘just happened?’


Greg J. Smith

A writer and cultural worker based in Hamilton, Canada, Greg is an editor for HOLO and his writing has appeared in publications including Creative Applications Network, Musicworks, and Back Office. He is also a PhD candidate within the Department of Communication Studies and Multimedia at McMaster University, where he is researching the emergence of the programmable drum machine in the early 1980s.

$40 USD