Hardware Physical Modelling


Anyone here using the Nord Drum 3 for physical modeling (sequenced or played)?


@kilchhofer Where in the path do you place the d0? I have a small 3U dedicate setup for this purpose, and I added a meager 2hp Comb at some point, which is a very fast delay (more or less, but you know that already), but I’m unsure of where to place it in the path exciter > resonant filter.

The Serge EQ brings some extremely realistic overblowing elements in the patch. That’s true. But you mention tracking … are you sure the body of an instrument is to remain untracked? Since the body length changes, by ways of the mechanics, I assume it’s also that body that needs to track the pitch change, at least somewhat.

I’ve had some really good brass results via various methods. One that works rather nicely is sawtooth in granular environment, believe it or not.

Check this video out at around the 3min mark https://www.youtube.com/watch?v=SwXMGWVD094

I’m also considering using pitch-tracking samples as exciters.


usually i patch it like this:

the 2hp comb doesn’t have an open feedback path, so i don’t think it is suited for this arrangement.
i use the res eq or the c0 as body filter, this doesn’t track 1v/oct, i see it like a fixed room, sure you can experiment with it, feed it back into the path etc, but i like it to be “stiff”, it sounds convincing to me (if that’s a goal anyway). granular methods at the exciter stage are wonderful, madrona labs kaivo works like this and you can create great dynamic blowing or striking variations. i have no desire to really replicate convincing instruments with this, it’s just immense fun to play around with these building blocks, especially if you introduce vca’s and envelopes and forget any “rules”, feed back everything into everything. the filter in my drawing sets the register of the instrument, i would set the d0 to the fastest setting possible which would be an soprano piccolo or something anoyingly high and use the filter as a octave divider, i still don’t know why it works this way, but it does, the overblow effects also happen at this filter stage, the more i open the cutoff the higher the sound but it jumps in these very natural sounding steps, dunno why.


So filter is in the feedback path of the d0. Got it, thank you very much for sharing this concept.

I was thinking about this when you posted: making use of two small (2hp) mixers to replicate the effect of a matrix mixer, as per the discussion above. Especially since the Resontant EQ already has a 2in:1out (odd/even outputs notwithstanding) mixer onboard.

That would open up the Comb. I used to have this and a LPF in the path of the feedback of the ResEQ going back to itself. I added the Clouds to pitchshift the feedback output and then went downhill from there by adding sampling for exciter. I never thought of using the granular to process the exciter, I’ll give it a try.

Long story short, I bought a small 4hp breath controller module (BC from Pulplogic) and that got me thinking about using that to control a modular physical modeling instrument. Totally worth looking into if you’re into that kind of control. Totally blows (pun intended) any other method of controlling the VCAs.

For anyone else interested my 3U system consists of the following ingredients, which I mix and match as I see fit: LPF, Comb filter, resonant EQ, mixer, granular processor, digital oscillator, sampler, VCAs, breath controller.


Is anybody here using a Rossum Morpheus in your physical modelling patches? It seems like it would be really great for a feedback filter or a body filter, and you could do a lot of awesome stuff by sequencing presets. I haven’t been able to find any examples of it though.


No but I’m in complete agreement. I think it would function beautifully.


@kilchhofer yesterday I tried your delay>filter feedback path idea using a 2hp LPF and a 2hp Comb (which can be considered a filter or as a fast delay line). I also used a 2hp Mix to create a pseudo-feedback path on the Comb. It worked rather nicely I have to say!

I passed everything through a Serge Resonant EQ as a body, as per your suggestion and it really really makes a huge difference. I’ve been using the ResEQ in a completely different way so far, but I now see what you mean by your post above. Changing the Comb resonance and filter (it has it’s own dumping filter in the internal feedback path) and then reshaping the sound via the Serge I went from low baritone sax sounds to trumpet brilliance without repatching anything.

It takes a while to find the right parameters in order to tame the feedback path, but it works like a charm.

I’m upgrading this small case to a 2hp MMF, a Plaits (as exciter), and a couple 2hp Mix modules more, just to be able to work with feedback better. Thank you for all the suggestions so far.

PS: the reason why I’m focusing on 2hp modules is because this is a very small 85hp 3U system and opting for minuscule modules allows me to retain the focused size but expand the path options. I always liked physical modelling but working this way, is a whole new world for me and it’s quite fascinating.


glad to hear that @ParanormalPatroler!
the steampipe reaktor ensemble helped me create these d0 patches, it’s quite simple in a way, but it sounds good and lots of variations are possible, here’s the core structure of the pipe:

right now i am experimenting with the exciter signal or the embouchure controls, i want to have more variations in how the sound starts, as we all know, the first few milliseconds basically tell what instrument it is. so to get some growl or stutter etc it’s really cool to add a vca+env combo into the mix and feed different signals to places for a short period at the start of the sound. the envelope shape is also crucial for this. something like this:


obviously if one takes a sample as exciter signal, or a granular controlled sample then this isn’t so tedious:) these patches are so extremely delicate a dynamic way to control them is necessary, pressure control or breath control etc.


Yes, that’s why I added the 2hp Play there. I plug a copy of the BC module’s envelope output to the trigger of the 2hp Play and it fires the sample once which will be helpful for “sound start” samples. One other reason for having the sampler there is to add samples of continuous exciters or even actual samples of the instrument, and then have it play through filters, delay lines, resonators etc.

I don’t have an LFO in there but I have some ideas. It can be funny how I need to completely change the way I’m playing the BC module in order to get the right sound for each instrument. Going from sax to trumpet meant I had to change sensitivity/offset parameters on BC and change my blowing technique, otherwise it didn’t sound right. I knew that would be the case, but it’s still a testament to how important the main envelope controlling the various parameters is.

I wonder if having the breath controller output control the various parameters (volume, filter cut-offs etc) with the same amount, is the right way to go. I’ve been experimenting with it and I haven’t come to a conclusion yet.

Also, @kilchhofer I’d like your opinion on exciters you’ve been using. I tried Braid’s sawtooth-throuhg-comb-filter mode yesterday and using your feedback path I really got close to trumpet sounds, which is to be expected as sawtooth is ideal for trumpet.

I see noise on your diagram above. Any other suggestions? I’m unsure what to go for when it comes to samples > granular or even waveforms > granular.

Thanks for all the insight. I’ll be posting videos with results soon enough.


for me the exciter signal isn’t really that important, at least not a sound signal, more a shaped dc signal. the fun part at least for me is to not use oscillators or signal generators except maybe noise but even that i found not to be that important. the more i look into well modelled instruments the more dizzy i get, my approach is really extremely basic and naive, or more experimental if you will haha:)
i try to split the instrument into two parts, embouchure and bore and in a way treat them almost the same eg using delay, filter and feedback but force the first into the second and have the well tuned pitch control only on the bore part. it’s hard to explain and based on trial and error. waveshaping i find important and offsetting signals, also inverting signals. with feedback such “simple” steps can lead to really big differences in sound. breath controllers in my case control loudness and brightness but that also isn’t a good solution, often i have hybrid wind instruments where in a lower register it sounds flute-ish and on higher registers it becomes reed-ish. i tried to maintain the same timbre over a bigger span with bandpass filters but man, it really is crazy complicated in the end, my whole case for a rather demented sounding flute which i also could carve from a carrot … but fun it is!


I’ve also found the most important part of the exciter to be its dc component. Just changing the the shape and amplitude of that has really big impacts on the tone of the entire model. I usually mix that with bandpassed noise with a lot of resonance.

I’ve been wondering about Xaoc Zadar as an exciter. I’m going to do some experiments tonight with ringmodding envelopes and audio rate LFO’s together to get some of shapes similar to Zadar’s more complex offerings, but the whole interface and range of modulation options on that module could be really great for this. Especially with its precision control over the duration of the impulse. I’ve found string models in particular to be really picky about how long the exciter impulse takes, with just a few too many milliseconds seeming to choke the patch and not let things resonate quite right.


ah wow haven’t heard of the zadar, thanks for bringing that up!
indeed it looks like it could be a very useful envelope generator for this task.


A friend brought it in one of meetups and, if I recall correctly, used it to open up a filter. That patch alone created some really nice sounds, reminiscent of modelling. I never realized an envelope would work that way, but on second thought it made absolute sense. I’m unsure whether Zadar allows you to load your own envelopes (and I wouldn’t know where to start finding the right ones) but it’s definitely something that can be used in physical modeling.

@smbols @kilchhofer what do you mean by the dc component of the exciter? You’ve both mentioned it and I’m unclear on what you mean. Is it shaping the audio using an envelope or, as I’m guessing from what you’ve written above, mixing in an actual envelope as exciter?


It doesn’t, and won’t be supported in the future either.


Thanks for letting us know. On the upside, it seems you can really shape the ones that are already there. I wonder if I can cover similar territory with the Omnimod.


Yep, given there are a lot of shapes in there, and you have a fair bit of control over them, i don’t see it as a reason not to buy one. I’m certainly planning on adding one to my rack.


it depends on the patch. in my case normally i use my pressure or breath cv to open a vca in the delays feedback path, this then let’s the thing oscillate. to then further shape the feedback signal i add or subtract a static or enveloped bias to the signal. in the end there’s no exciter signal but control over the oscillation and offsetting or waveshaping this feedback signal. if you look again at the reaktor screenshot i posted above, you see two feedback signals coming back from the filter into the delay line, one called feedback which is basically an amount control and the other called push-pull. this is how the push pull macro looks inside:

push adds a static voltage (0-2), offset shifts it from -1 to 1. input B is the feedback signal, input A the exciter which is a mixture of pure envelope and noise. these controls interact with each other quite drastically, the feedback amount control has to be adjusted constantly, if there’s a waveshaping process in the path then this makes it more delicate, wavefolders for example usually add a dc offset to signal anyway and i had good results with only a wavefolder in the path to shape the signal.
the envelopes mentioned can be inserted as dc offset control, waveshape amount, generel vca control or even at the filter stage. it really isn’t a fixed recipe and alot of trial and error but the “normal” way, of exciter into resonator like elements for example i usually find quite boring.

now this is obviously way easier in reaktor but i patch it similarly with my modular, the important thing is that the delay must handle dc signals, otherwise this doesn’t work.


also i have to add i speak only about wind type sounds, not plucked string instruments, for such sounds the shaped noise into resonator method works pretty well.


I do something similar to kilchhofer, but for strings I usually mix filtered noise with a +5V offset, then VCA that mix with a very quick envelope to act as the exciter. Or I’ll envelope filtered noise on its own, and mix that in with a shorter burst of DC voltage like a trigger pulse that fires off along with the envelope. Just sending envelopes straight into the resonator works too, but usually leads to a thin sound. Adding the filtered noise seems most useful for filling out the low end.

I messed around with some envelope shapes inspired by Zadar last night and got good results. I use a Sin Phi Miasma for envelopes, which is a Befaco Rampage with extra goodies, including a 4 quadrant multiplier output, so using that with one channel set to chaotic, audio rate oscillation and the other to a quick envelope gave interesting results. (I didn’t test out how similar this is to just VCAing a regular oscillator to act as an exciter but I imagine it would be similar since that’s essentially what the 4 quadrant multiplier is doing here.)

Adjusting the tilt of the oscillator portion, similar to how Zadar can stretch out and skew its envelope shapes, has a very noticeable effect on the tone of the resulting sound, similar to plucking a string in different places along its length. It’d be great to control that with a joystick or something to add expression to your notes.

Unrelated, but last night I also found that for string patches, adding a little bit of self-FM from the filter in your feedback path sounds great and makes the string seem a bit more natural. You can push the feedback a little harder, too since it can’t run away as easily with the filter frequency moving around.


Now this is some hardware physical modeling: