[ann] ODE integration as synthesis engine, SMC 2021 papers

Hello all,

SMC Torino, like 2020, happened in an online format, thanks to Andrea Valle and the organizing team! The proceedings contain a paper about miSCellaneous_lib’s Fb1_ODE interface:

Also, I would like to point to the work of Dario Sanfilippo, who implemented constrained differential equations in Faust:

My last piece exclusively uses a specific non-autonomous equation type – here’s a short sound example:

https://www.daniel-mayer.at/werke/Matters_7_en.htm

A bit of documentation and description of the equation (written in the middle of the experimentation phase, paragraph “Snapshot: Current Synthesis Experiments”):

https://www.researchcatalogue.net/view/921059/922503

Have fun, regards

Daniel

5 Likes

This looks amazing. I look forward to reading this and Dario’s papers. My paper next week at the AI Music Conference also has a massive shout-out to both yours (discussing synthesis design in the context of human-in-the-loop interaction, extending the metaphor from synthesis to neural network training) and Dario’s excellent work (in which I further modify his modified Lotka Voltera model for even more tomfoolery).

I will be interested to implement one of my crossfeedback or diff eq models in Fb1. All my stuff is faust compiled to SC at the moment. Daniel, do you have a sense of an efficiency comparison between the two approaches?

Thanks for the hint, I wasn’t aware of all program points!

No, I haven’t. I wouldn’t be surprised if Faust compiled to SC is eating less CPU. However, cross-feedback (and -feedforward) can be tightly formulated in Fb1 as everything can be multichannel. It’s just that one has to be careful with indices. For this, using meaningful variable names inside the Fb1 Function can help (an example from a recent thread: Phase Distortion with Fb1 - #4 by dkmayer, for cross-feedforward you could, e.g., use stereo sources and then swap the third indices of in, analogously for cross-feedback the second indices of out, outSize would also have to be set to 2)

@dkmayer very interesting article :slight_smile: do you have a code example for the Buffer rewriting technique? Im still searching for synthesis techniques which avoid these well established cliches thanks alot :slight_smile:

I noticed with delay, all video presentations are on YouTube now:

Dario’s and mine:

4 Likes

Update, the above link is outdated, the SMC paper on Fb1_ODE is located here:

the whole proceedings: