Traveling Boyfriend is a project I've been working on with Shelby Mass for a few years, which mixes folk, electronic, and historical music. Shelby's songs evoke the symbolism of mythology and dreams, to which I add found sound, percussion, and synthesizers.

Death of a Flower is planned as our first EP, personifying the last flower of the summer drowning herself to allow winter to come. Here are some demos, in various stages of development.

girlhood-10-5-25.mp3

inside-a-summer-flower-10-18-25.mp3

cosmicomics-fish-song.mp3

wyrbtw-10-25-25.mp3

molds-pollens-worms.mp3

Early in 2025 I started playing Animal Well on Nintendo Switch (highly recommend this game for anyone who likes really challenging puzzles and secrets). This and other Metroidvanias like Hollow Knight inspired me to try to make a little bird platformer game in Unity where you play as a kiwi, solving puzzles and finding secret areas.

There's some character physics, very basic NPC and object interactions, some portals, a weather system, and a camera system based on which room you're in, like Animal Well.

I also threw together a pathfinding system where the agent simulates a series of actions against a copy of the world's physics. Unfortunately it lags pretty heavily because it's very unoptimized and I wasn't able to get it to run in a background thread.

Anyway, I still don't know much about game development and I may never finish this, but it was fun to learn.

AM synthesis with intaudio

intaudio is a Javascript library for working with audio interactively in the browser, ideally enabling limitless signal processing and experimental musical instrument design on the Web.

intaudio is declarative, so it lets you "connect" arbitrary components together, similar to how one might connect objects in Max/MSP. The Web Audio API (which this library uses, of course), is designed with this same general idea, but it only has a few top-level features such as oscillators and reverb, and it can be pretty unwieldy for complicated signal processing.

The main totally-complete feature at this point that, in my opinion, makes this library really powerful is the ability to perform arbitrary sample-rate operations between signals.

FM synthesis looks like this:

You can even have a sine wave control the L/R balance of a signal:
(I'm thinking about creating a utility like a signal.pan(...) method that would make this easier.)

One more example, which I think is really cool--a custom echo effect where the delay value is controlled by an oscillator.

The endgame / wishlist of this library would unify audio signals, "event-based" data (slider controls), and STFT data seamlessly like the following:

But this doesn't work yet!

bells and machines was planned as a second kmodp album in early 2025. It was about 70% finished at the end of 2024, when suddenly my hard drive failed irreperably and I lost all files from this project as well as all other projects, besides those I was either storing on Google Drive or had backed up previously.

Thanks to some random bounces I was able to have a full (and relatively complete) version of lifelong life, which I released in 2025, and all other tracks remained only as short portions or early versions of bounces.

Here is broken machines, AKA what I have left of that album as well as what I'm doing to reimagine some of these recovered or lost sounds. Some of these are partial bounces while others are new tracks created in the same vein or from these bounces.

m-guit.mp3

bell_loop_with_june.mp3

fulcrum-1-v1.mp3

fulcrum-1-v2.mp3

dark-fulcrum-2.mp3

fulcrum-3.mp3

I am experimenting with layering spoken voices for a few tracks, using samples from the Mozilla Common Voice dataset and public domain audiobooks from Librivox. I layer and distort these voices to cause dissociation from the content of any one narrative, as if one is trying to consume multiple types of information at once or has fallen asleep while listening to spoken voices.

To me this is a weird combination of unnerving and calming. Growing up as a pretty shy child who was exhausted by social situations like holiday parties or family visits, I have memories like this, being half-asleep in the corner of a room while people were talking around me.

dont-touch-me-4-19-25.mp3

ssommeil-9-7-25.mp3

anything-i-do.mp3

I came up with a track called zero-nine consisting of repeating commands from an audio speech commands dataset over a slow rhythmic backdrop made from sliced field recordings. This was a full track, but unfortunately it was completely lost except for this segment that I bounced at some point.

zero-nine.mp3

I played around with these ideas in a few other unfinished tracks as well.

fast-loop.mp3

tremble-plus-pécheur-9-7-25.mp3

This was going to be the last track on bells and machines, and I was able to find an (almost) full bounce of it.

I used a simple synth patch without velocity control and a fixed set of voices that cut out when new ones are introduced. With the lack of expression of the instrument and the meandering rhythms, I feel that these ideas could go on forever, satisfied with simply chasing their own tails.

closing.mp3