My experience at the International Conference for Digital Audio Effects (DAFx) 2023

A couple years ago, I started a project, DelayCat, as an undergrad under the advisement of a professor, Brian Hansen, at University of California, Santa Cruz. The project was meant as a technical learning experience for me, through realization of an idea I had floating through my head. However, a bit later into the process, we decided there was enough there to write a research paper on the topic. A year and a half later, the paper, Feature Based Delay Line Using Realtime Concatenative Synthesis, was accepted to DAFx and I found myself attending and presenting at the conference this past month.

As context for my experience, I am a recent graduate with a b.s. in computer science and a minor in electronic music (graduated right before the conference), which is important in understanding my conference experience, among a field of mostly masters / phd students, faculty, and company research teams – all very experienced individuals.

It was a wonderful enriching and educating experience for me, where I was able to soak up information, learning and being taken into the community of this field.

Without getting into the gritty details, I am going to talk about some of the things I learned from my experience.

About the field

The work presented mostly fit into the following main categories: Physical Modeling, Virtual Analog (VA) Modeling, and Machine Learning (ML) AI, as well as some smaller miscellaneous other categories.

In particular, as the current craze throughout all of technology, ML dominated the conference. There were many different ways ML was being used, but primarily it was being used to create blackbox or greybox virtual analog models and emulations of a range of different analog effects. Company research teams, especially, were interested in this, presumably because of the money it presents through relatively easy / quick results and industry buzz. In this way, I think the ML work can be thought of as a category of its own, but it also as an application of ML towards the other categories (i.e. “using ML to do VA” or “using ML for physical modeling”), though there was also some less categorically grounded ML work.

Throughout the rest of the work, there is too much to talk about to break it down, but I can give the following generalizations:

Physical modeling was the more physics / math side of the spectrum, used to make software models of existing (or imagined) physical instruments / systems to create timbre.

Virtual Analog Modeling was the electrical engineering realm, working on translating the behavior of electric circuits directly into software, to create an array of different effects based on real analog hardware.

The other categories contained work related to reverb, spatialization, perception, analysis, and synthesis, to name a few.

About my work

My work stood out a little bit from much of the field, as something a little bit different. This could be attributed to several different things: my ducking of the current ML wave, instead focusing on slightly older methods, my background outside of an established lab or company research team, or a number of different things. But one way or another, it was a bit different.

I believe this is part of what made my work interesting and valuable, and I take pride in this, however it also illuminated some of its deficiencies, and some of my own.

Namely, I think a lot of this came down to lack of experience (due to young age) and lower academic / standing, something which I feel very confidently about my ability to grow through – experience will come with age, and I have proved to myself my ability in the academic world.

But it also illuminated a slight lack of attention / value given to the user experience and visual quality, and some blindspots to my knowledge of the field.

Even for this crowd, it took a bit longer than I expected to get my idea / architecture across, reminding me once again of the user experience problem involved with this project due to its complexity and novelty of architecture, a problem which I still hope to solve to a greater extent than I currently have. The many rough edges still in the plugin interface, definitely didn’t help in this regard as well. Though, I will say, I don't think this is by any means a trivial problem or a total oversight on my part. Rather, I think this was a reinforcement for me and more information on the issue.

Regardless of that, people were intrigued by the system and generally impressed by the sonic results, which I count as a win. And I received valuable feedback through discussion with my peers.

Furthermore, getting my work out there, having eyes on it, people playing with it, was valuable for me in a more nebulous way through its impact on me. A reminder to relinquish control, and to let my work live and breath, participating in the magic between the lines that happens between different people.

About my future

Compared to the conference crowd, generally speaking, my strengths lie in my computer science / software engineering skills, as well as (I believe) my creativity and (depending only my goals) background in game design and development, and not as much the complex math, physics, and electrical engineering (at least at this point).

Furthermore, I currently lack extensive training in the buzzing ML AI world and have a general skepticism for its overblown use, though I am currently training myself in this area, due to the overwhelming demand for the skills, and the fact that I don't feel like I can harbor the skepticism in good consciousness without properly (thoroughly) understanding and contextualizing for myself.

With that in mind, the main options ahead of me seem to be: focus on software, or go back to school to pursue research. Audio software positions, especially those overlapping with my experience in gaming, are quite limited, but I believe in my abilities if and when I get my chance. And if I decide I want to pursue research, or that I need more schooling, I would be excited to apply myself to that route as well.

The conference also got me interested (through some keynote speakers) in working on hearing aids. This is an interesting application of my skills through which I could more tangibly help people. Though this is an even more narrow section of the field.

There are many exciting things to work on.

Afterword

I just want to take this chance to encourage anyone who has something they are interested in working on to just dive in! Don’t feel afraid to see if others are interested in collaborating or helping as well, be it faculty if you are a student, or peers, or friends. In fact, feel free to ask me if you find yourself in that position!