Hi all,

I have Andy Farnell's book Designing Sound, it's fantastic and it taught me a lot.

I still feel like I would have trouble modeling an arbitrary sound, and I'm not super well versed in physics -- is this a problem? I was wondering if there were any other online or book resources for learning about physical modeling in the context of Pd, or any other dataflow language.

Thanks

JOS is the king of physical modeling (mostly waveguides and such, he started with Karplus-Strong and went from there). I get the impression that Farnell is more of the modular synth programmer/foley artist school of sound design, i.e. making something that "suggests" a specific acoustic sound. JOS is totally rigorous, like a mathematician. He starts with the physics of it and only suggests practical concessions at the end. The nice part of this is that the results are really, really good, BUT it has to be something that can actually be modeled reasonably well. A lot of stuff that's essentially trivial in the physical world (throwing your car keys in the air or draining a sink full of water) ends up being way too difficult to model accurately, so this approach mostly sticks to the known universe of musical instruments.

https://ccrma.stanford.edu/~jos/

It's not geared toward Pd, but a lot of it can be implemented without too much trouble. See his earlier and simpler waveguide stuff for a reasonable introduction.

Hey,

I don't know this book, but I know about physical modeling and that is really easy in Pd and it's not necessary to know much about physics - it's more a trial and error thing.
I'd advice you to inform yourself about using the PMPD library. It's mainly about mass-spring-damping waveguide meshes of different dimensions and their computation.
Another form of physical modeling synthesis I really like is Modal Synthesis in which you have a sound source like a transient of white noise or a pulse train exciting a delayed feedback-filter-network or a recursive filterbank...should be quite easy to model convincing natural sounds.

best

Elden

my 2 cents, is that sometimes it can help to get a recording of the actual sound you want to model, and open that up in a wave editor. You can figure a few things out by looking at the waveform.

Thanks for the help everyone!

@acreil I will look into Jos, to be honest his work seems a little above my head atm, but I'm sure I can extract something from one of the papers.

@elden I will look into Modal Synthesis. On a simpler level I've been using white noise as kind of a block which I can subtract from to produce something natural sounding.

@mod Thanks, I should look into this. I could see how a spectrum analysis would be really useful for reproducing odd harmonics.

@isBanekin said:

On a simpler level I've been using white noise as kind of a block which I can subtract from to produce something natural sounding.

Yes, that would be a method (Spectral Synthesis), but you won't get the "special something" of natural sounds in which some frequencies decay faster and others decay slower (not using envelopes). You could use bandpass filters or graphical EQs in a slightly delayed and volume-adjustable feedback to model that - in fact that's Modal Synthesis in simple.
You could check out NI's Prism for Reaktor!

I did my final year project for my degree on physical modeling before this summer. I chose the guitar, specifically electric, and I built my final synth in PureData. Is this what you mean? I can upload my final annotated patch if you want?
I read Julius smith's entire thing on his site, and it took a while to work out how to implement it in pd, especially with delay lines. Built one in Matlab first, then did it in pd because I hate programming in text like with C++. Built a delay with feedback, bit-crusher, analogue/digital overdrive and distortion, as well as ring mod and amplifier feedback loop. I ran out of time, so I didn't get to do pickup design-specific modeling, or sympathetic resonance. I did manage to get different types of excitation (pick or finger etc.)
I've got a pdf of my dissertation hanging around somewhere as well, puts all the physics stuff into laymen/musician's terms
anyway, let me know if this is what you're on about

@elden

Appreciate the clarification -- so if I understand you correctly I can use Modal Synthesis by staggering the change in noise generators running through a bp~, in volume and in center frequency? Do you have a patch I can look at?

@littleredcourgette

I'd love to see it, that is completely related to what I'm doing, thanks.