What did I learn?

Technologically: My first pass at this project was also my first experience coding in C# and dealing with a number of HCI and computer graphics concepts like shaders and interfaces. It was an opportunity to get hands-on with the interwoven systems of interaction design, testing how frontend and backend decisions affect the emotional valence of the resulting experience. And of course, I dove down the sound spatialization rabbit hole, learning about HRTFs, ambisonic file encoding, spatial perception, and spatial DAWs. When I returned to this project earlier this year, I felt that I could approach these topics with more confidence.

Musically: My composition was heavily inspired by Joseph Schwantner’s …and the mountains rising nowhere and John Luther Adams’ Become Ocean, works whose attention to instrument pairings and texture pays off on grand emotional scales. I was ambitious with my use of extended technique, asking the brass to blow bubbles with their mouthpieces, the low strings to rustle their instruments, and the whole orchestra to vocalize as part of the overall texture. There’s also a structured group improvisation, where I play with agency as it affects density and time. I spent extra time recording demo videos and working with individual YSO members in order to iron out the performance of these experimental elements. I also worked with the recording engineer at the Yale School of Music, figuring out how we could modify the standard orchestral microphone rig in order for our resulting stems to work in a spatial multichannel mix.

Personally: This project put me in charge of a yearlong collaboration between the CCAM, the Yale Music Department, and the Yale Symphony Orchestra. I practiced my communication skills constantly, resolving conflicts related to schedules, budgets, logistics, and policy. It was challenging and rewarding to work with so many people in service of an overarching vision, to methodically plan and execute a project with so many different moving parts. It was also an important exercise in adapting to feedback from my mentors and my listeners, putting extra effort into the elements of my work that resonated most with my community.

What did I learn?

Technologically: This turned out to be a pretty complicated patch, and the visual coding environment calls for a high standard of organization. I approached it as an exercise in abstraction and consistency. One fun problem I ran across was how to create a nested for-loop that updates in real time. Max doesn’t have built-in loop functionalities, so I ended up using two ticks on different timescales: the render tick, which happens on every frame of the simulation, and a metronome tick, which happens at regular intervals as a function of Max’s built–in millisecond clock. I realized that playing with the render speed and metronome speed gave some interesting and unexpected results, so I decided to include control over these values as part of the user interface!

Musically: A lot of interesting new music deals with curating the unfamiliar, with taking comfortable structures and stretching them in unexpected directions. In designing this simulation to also work as an instrument, I had to think about the composers and performers who might want to use it, the questions they might want to answer with it: What if a musical octave were defined by some arbitrary interval? What if our chromatic universe had an arbitrarily high number of pitch-classes to choose from, but our scales still had seven notes? How can rhythm and harmony serve as forces of constructive and destructive interference for one another? Conversely, I had to think a lot about what restrictions to impose in order to make the interface feel expressive.

Personally: I find the politics of FiPS interesting. Much of the literature centers on analyzing or generating Western music, on a special relationship between Western art music and values of symmetry and order. I wanted to create this simulation to showcase just how broad this generative space actually is, how much unfamiliar, dissonant, wacky music can come from the exact same pursuit of symmetry. In general, I think interactive experiences like this can do much to separate the idea of ‘we can generate structures by X algorithm’ from ‘the values underlying X algorithm also underlie the structures it is able to generate’.

What did I learn?

Technologically: I wrote nearly all the code for this project, meaning that I spent over a year studying Bayesian statistics, principles of machine learning, and relevant libraries python and MatLab. It is hard to overstate just how much I read, learned, and practiced in order to bring this research to life, and although I have much more growing to do as a programmer and engineer, I’m incredibly proud of how my efforts turned out.

Musically: I also contributed much of the introduction and discussion text for this project, drawing on my prior studies at the Yale Music Department to compile a musicology reading list for my collaborators. It was an interesting and difficult task to draw from some very humanities-flavored disciplines in the service of a very straightforward, STEM-flavored research project. It led to interesting discussions on how we want to define music, how we relate our specific symbolic dataset to other cultural artifacts, and how we want to delineate the boundaries of the evolutionary analog between biological and musical structures.

Personally: This was my first major research project after college, and it served to instill my passion for research. Both of my collaborators moved on from the lab before the paper was finished, and so I stepped into a leadership position in order to meet our revision and resubmission requirements on time. It was a rewarding process to work in that dynamic: to ask questions and learn from people much more experienced than me, but also to find the confidence to contribute my own voice and expertise. By the time we published our final manuscript, I felt that I had matured significantly as both a scientist and an academic collaborator.