Physical Modeling

interesting stuff

This page is the highlight for me, although I am not actually sure the exact method they use to generate the instruments. Upon reading the site over again, the method they use may be similar to other Phys Mod algorithms (seems like they code the shapes, not based on models?)

Anyhow, cool samples:

http://www.ness-music.eu/target-systems/modular-environments

The one long form piece they have available is good too

http://www.ness-music.eu/music/ashes-to-ashes

Oh wait, I guess all shapes are code. it's just a matter of the way they are generated (gui with underlying code vs. coding directly) and the file format/programming framework, yeah?

I don't know anything about doing Phys Mod from code or otherwise engineering it. Kaivo is disappointing so far because the physical models are all preset, don't know how I would modify them

  On 9/19/2015 at 1:32 AM, ganus said:

RE: using 3d models to generate sound, pretty sure it's been possible for a while, they did some of that here with interesting results:

http://www.ness-music.eu/overview/physical-modeling-synthesis

 

Good site to explore, although I wish they would share some source code. Shows some interesting possibilities

I'm having trouble finding anything about it online, but topographical synthesis* is kind of related to this. Waveforms generated by tracing a path across a 3 dimensional surface. IIRC the standard method for it is to have a point moving in a circle on the surface and generating the waveform from its Z position. The X and Y coordinates of the center of the orbit can be modulated to change the shape, and the speed of the orbit changes the frequency (one revolution = one cycle). It's an interesting idea but I can't say I've knowingly heard anything made that way.

*it's described in The Computer Music Tutorial but I don't think it really caught on outside of 90s academic music

  On 9/19/2015 at 2:52 AM, ganus said:

The one long form piece they have available is good too

http://www.ness-music.eu/music/ashes-to-ashes

yeah thats fucking awesome, this is actually very very similar to the stuff i've been working on for the last year, although this is a tad more minimal. really interesting to hear something like this (and it was made in 2008? i need to play some serious catchup!). thanks for finding/posting. This sounds like straight up Vl1 to me, the only difference is maybe the spatial placement, that sounds a little like the models emit stereo sound, which is something I've always thought would improve a physical model's sound.. so im very intrigued that this can all be done in Matlab(?) thats really surprising. I need to try and work with this group and their models, time to brush up on my academic sales pitch lingo.

https://ccrma.stanford.edu/~rmichon/faustSTK/

This is a bunch of implementations of physmod algorithms in Faust, which is a DSP programming language that compiles to C++ (but there are also converters for puredata, max, vst, ladspa etc..). Now the cool thing is that these are all based on MusicKit and SynthBuilder algorithms, which were in turn part of the Sondius program at Stanford. The Sondius project was at some point licensed by Yamaha (in the form of a patent package as far as I understand it) and AFAIK this was used as a basis for their physical modeling synths. So this might be coming quite close to a software version of the VL1, maybe.

I haven't tried to actually run any of this though lol.

https://ccrma.stanford.edu/~rmichon/publications/doc/DAFx11-Faust-STK.pdf

Paper on the faust-stk project with details on the algo's.

It's unfortunate that the audio examples at the bottom make it sound like a Soundblaster AWE demo from the mid-90s

lol

There's some better demos of synthbuilder here

http://scandalis.com/jarrah/PhysicalModels/index.html

Mostly a matter of taking the time to add all sorts of expression parameters, I'd think.

Reading that pdf now, it seems they just assigned ADSRs to things like breath power and stuff.

Model this:

Well I finally managed to install Yamaha S-YXG100 PVL inside a VM with (almost) playable latency. Here's some initial noodlings. Until I manage to get a real VL machine, this will have to do...

These are the Bamboo and Shakuhachi presets mostly, with some Voxasax and Tuba near the end.