Hi! semi-n00b here. I'm working on a modular-synth-style virtual synth in pd. I'm using arduino to make a physical interface, similar to those on old modular analogue synths.

I would like to have a decent way to patch output from, say an oscillator, through a series of pd effects. However I want to be able to change the order of the effects in the sequence between the oscillator and dac~ on the fly. To make it more complicated, I intend to do this only with my arduino interface, no gui at all.

Is there any good strategy for patching effect modules together without either a gui or physical patch cables? Maybe something like a radio item that sends signals in numerical order?

Thanks for any advice, sorry if this is unclear!

Alex

It depends on your setup. If you have set number of slots, then it would be most efficient and practical to have each effect slot have each available effect loaded on start-up and simply turn on/off the effect you want with [switch~]. If you don't have a limit, then you would have to go with dynamic patching, which can lead to temporary drop-outs. There are ways to limit the drop-out time (and possibly eliminate if you're lucky), but it is risky for live performance.

didn't someone post a working system with dynamic send~ objects done with subpatches???

http://puredata.hurleur.com/sujet-5951-interchangeable-effects-throw-catch

umm, maybe shaun.pd's vapourware was the closest we got????

if you do it using message-based dynamic patching, then you definitely want to keep the dynamic stuff as simple as possible and in subpatches or abstractions.

Yeah I did this by re-routing [receive~] objects based on the previous fx output in the chain. It's possible without dynamic patching.

But I can't find the post I wrote

Just an input : i did a modular fx-chain in pd with dynamic patching (dynamic instantiation in empty subpatches, static patch cords) . I couldn't get rid of drop-outs, plus it's a real mess to manage (gotta use some tricks like the "dirty" flag).
I now try to use [mtx_*~], it's far more logical for the job since all we want is rearrange a route through a given number of units (hint : think matrix analog synths). It's also easier to store states in .mtx files.

[mtx_mul~] is nice for fixed routing, but in situations where you might want any number of effects, it can be limiting. The number one trick, I've found, for limiting drop-outs in dynamic patching is to turn DSP off, patch in the abstraction, then turn DSP back on. The reason being that the DSP graph is redrawn for each new DSP object that is added one at a time when it is turned on. If DSP is off, then it only re-graphs it when DSP is turned back on. Sometimes you can get lucky, and you won't get a glitch.

I won't trust dynamic patching in live performances, though. You're much better off setting limits and working with them.

I read something about Harrisson Mixbus (a modified version of Ardour) : they apply an "always on" policy when it comes about audio components. Isn't pd supposed to be a deterministic program ? You're right on this one : live patches should be deterministic too, so you do have limits. I even would be tempted not to use [switch~] to keep the cpu load even.