Very early stages but I'm excited about this: I'm developing a new kind of resonator. No delays, no Karplus-Strong, just a pure partial differential equation solver running in real time.
Comments
Log in with your Bluesky account to leave a comment
Why? Karplus-Strong from the outset was designed as a prudent optimization of a complicated scenario. In a lot of circumstances, a plucked string behaves much like a delay with feedback. This is not always the case though, for example, when material properties can change over space and time.
Famously, strings drop in pitch a bit a moment after they're plucked, because the tension from plucking the string affects its frequency. Different effects can be achieved by plucking a string with more than one finger, changing harmonic content. My intent is to simulate these behaviors (and more)
There's an incredibly diverse set of real-life systems that can be described by PDEs, which often do not have simplifications usable for audio. Many of them can be simulated using exactly the same solver. Thus far, I've implemented the Schrodinger equation, the Hodgkin-Huxley model, and Advection.
The end goal is to make it so that you can not only change the resonator's material properties, you can change the topology of the resonator on the fly, smoothly morphing between plates, tubes, strings, and more!
Basically yes, but I'm going down to the level of the wave equation itself, and I'm also implementing a few other PDE systems to simulate other kinds of nonlinear effects
This is a different method, but the connections between raytracing and the wave equation run deep. Ray tracing is sometimes used for designing reverbs as well!
UI is one of the more tedious parts of this, so I'm using the old skin while I flesh out this new system I'm working on. This is like a much more sophisticated successor to Watertable.
Very excited to try it out, it reminds me a bit of some of the Frequency Domain modules (Ball of Confusion, Box of Revelation, Dance this Mesh Around) by @almosteric.bsky.social
Definitely can understand UI can be a bit of a bear. Watertable has always been one of the most visually appealing to me.
Oh wow, neat, didn't realize you are the developer, that makes a lot of sense now! Excited to try it out when it's out! I actually played around with Watertable over a year ago to try and get a Karplus-Strong-like thing with the Runge Kutta model, and it worked, but also ate my CPU for breakfast 😅
Yeah, it's quite challenging to get this stuff running acceptably fast. It's roughly equivalent in computation to running the stock VCV VCF with 64 voice polyphony. I'm gonna experiment with some additional CPU-saving options on this one (which trades off for higher latency for example)
I definitely can understand, it's doing a lot of work in the background. I want to start doing smaller patches with modules that I haven't had the CPU for with my more orchestral-minded layered, maximalist VCV patches. It will force me to simplify composition & let the modules do most of the talking
Comments
Definitely can understand UI can be a bit of a bear. Watertable has always been one of the most visually appealing to me.