Mixed media and generative sound art, 2021
On exhibition in the Search Gallery at Science World BC
Excerpt from Sound of Tree Rings
The Sound of Tree Rings is a mixed media installation that includes a generative soundscape composed from the tree rings of old growth cedar ‘tree cookie’ in the Search Gallery at Science World BC. The project represents the tree’s lifetime of growth using sound. Software scans slivers of the tree rings from the centre to the edge. The bands of dark (high density wood) and light (low density wood) are imagined as different aspects of music, such as loudness, pitches, and intensity. In the gallery, you can hear the music swell and recede, matching the slower and faster growth of the tree through the seasons.
Interview with Simon Lysander Overstall, composer and computational artist for Sound of Tree Rings:
KYLA
Hi Simon, could you please tell us about how Sound of Tree Rings began?
SIMON
When I was down at Science World with Maria and Julie very early on in the process of the Wild Empathy project, they showed us this large tree cookie: a cross-section of a tree, that was from a Stanley Park blowdown during one of the storms in 2006 or 2012. After we had a look at the tree cookie I mused about how we could do a sonification of the tree rings and Julie was delighted by the idea. So I began working on a software program to do that and, at the same time, Science World began treating and cleaning up the tree cookie. You can see in the pictures where they filled in some of the decay, and it’s all cleaned up and has a finishing coat.
KYLA
In terms of the sound software you built, can you tell us about how each ring is being read?
SIMON
Part way through that cleanup process, when the tree cookie had been pretty well sanded but not filled or anything, one of our research assistants took a large number of photographs of it. We got a set of photographs of the entire face of the tree cookie and these were then stitched together to make a 16000 pixel square image (actual pixel dimensions are 16383×16383). I made the software with Max/MSP, a programming environment designed for creative media work and sound synthesis. The software reads that large, very detailed grayscale image of the tree cookie as a texture (a texture is an image that is on the GPU), and it takes a very small, rectangular sliver of that texture from the centre out to the edge. That sliver becomes the data that I use in the rest of the software to make a composition which happens over a length of time. To use the data in the composition the sliver is read from left to right over the duration of the composition. Left to right is from the centre of the tree to the edge—the direction of time in the tree’s lifespan—the tree was youngest in the centre and oldest at the edge. The software analyzes the grayscale values and these go up when it’s lighter and down when it’s darker. These changing values are used to control parameters of the sound synthesis in real time.
KYLA
And how would that translate into sound, what are the corresponding tones?
SIMON
To create the sound, I’m using a synthesis technique called physical modelling, and it’s similar to 3D modelling for visuals, which simulates how light will reflect off various shapes and so on. Physical modelling synthesis works by simulating how various shapes would vibrate and produce sound if struck, or rubbed, or whatever. Usually, physical models of versions of existing instruments are made. There are physical models for strings that could be bowed or plucked, and there are physical models for membranes that would be like drumheads, and models for other types of instruments too. You can go into a lot of detail trying to design instruments. People replicate things, but you can extend beyond what’s possible in the real world.
The idea I started with here was to use physical modelling to make some wooden discs that were somewhat like the tree cookie, or representative of the tree cookie, and then use the tree ring data as something that would activate them—make them resonate. In the end I developed that a bit further to make something like a stringed instrument or string bass. The data bows the strings, with other parameters being controlled and adjusted too. A snippet of the data becomes a shape of how the bow moves across the string, and another part of the data increases the probability—the chance—that the instrument will be bowed at that particular moment. The software keeps asking, “Do you want to bow now?” And the answer is more likely to be yes when it sees a light part of the slice. Overall, it’s more active in light coloured portions of the tree rings and less active in darker ones. In the software there are six one string instruments playing, each in a slightly different pitch range, though all quite low. The end result is a very ambient string sextet with a different chord each time, and this works quite nicely in the exhibition space. Once the software reads through the sliver, it finishes the composition. Then it spins that image of the tree cookie, and takes a new sliver. It keeps doing that indefinitely.
KYLA
How is it installed in Science World, in the exhibition space?
SIMON
There are two speakers placed above the cardboard cedar branch cut-outs we installed in the ceiling beside the hollow tree, and the tree cookie is placed underneath them. The computer is installed out of sight and there’s no screen or image of the software. Because the speakers are installed above the branches, so you don’t see them, but definitely hear them. When you sit on the tree cookie the soundscape comes down from above.
KYLA
Do you feel like there were any surprises in the process?
SIMON
I had a number of frustrations for sure. The physical modelling is interesting and a great sounding synthesis to use but it can be volatile because if you break too many rules of the simulated physics, it can sound horrible, or not at all. Similar to real world instrument making, if the wood is too thick then it’s hard to make it vibrate and sound good. If a string is too long to play a high pitch it screeches. And, physical modelling uses some very technical parameters. There’s one called the ‘poisson’ ratio, which refers to when you apply a force to something, how it actually bulges out, or how much it compresses. I wanted to simulate the material of the tree cookie so I researched material properties of cedar wood information and brought it into the software but had to massage the numbers to come up with something that worked.
It also turns out that having actual hands for bowing an instrument is helpful. With a little practice it’s fairly easy for human beings. But even amongst humans, if you’re new to it, when you bow a violin it’ll likely sound horrible. It turns out that the software, scraping a bunch of arbitrary data from an image of a tree cookie, is not very good at bowing. So I had to play around with things and figure out how to nuance the data into the right range so it didn’t screech and honk, pierce ears and blow out eardrums while it was bowing an instrument. This was important for the Science World install, where there’s a general audience with lots of kids.
KYLA
When I listen to the sound it reminds me of some experimental music. How does it loop at Science World?
SIMON
I’ve set each sliver reading and composition to be about 60 minutes. When it’s done there’s a break and then it just restarts by rotating the image and choosing a new sliver. So it doesn’t loop per se, but the software composes a new piece based on another sliver, and each time it’s a bit different. This process can go on indefinitely and Sound of Tree Rings creates a new experience each time it’s listened to.
