Physical modelling drum synth?

[modey]

Alright, so I love EZdrummer and BFD and all that sample-based stuff, and they're really great for acoustic drums, but do any decent physical modelled drum VSTs exist? I'm talking realistic stuff here, I know it's probably not easy but I figured there must be something out there, right? It'd be nice to be able to fine-tune the parameters of a simulated acoustic snare, for example.

[Guest]

Haven't used his plugins a lot, but xoxos has a bunch of free modeling vsti's here:

 

http://www.xoxos.net/vst/vst.html

 

can take up some cpu though and you probably need to stack some other effects on top for more oomph.

[th555]

What was that melodic percussion vst that was posted here some time ago? It had this "unique" feature of coupling the vibrations of two separate elements. IIRC it did a lot of regular drum sounds as well.

 

Edit: Chromaphone it was. I've still got the demo lying around somewhere.

[kausto]

IL Drumaxx

[mcbpete]

What was that melodic percussion vst that was posted here some time ago? It had this "unique" feature of coupling the vibrations of two separate elements. IIRC it did a lot of regular drum sounds as well.

 

Edit: Chromaphone it was. I've still got the demo lying around somewhere.

Yeah Chromaphone is great, the demo that sold it to me was in fact the drum one -

 

http://s3.amazonaws.com/aas_sound/Chromaphone/DJChampion-Grandma.mp3

[AAAAh]

i was gonna say xoxos pugins

[th555]

Thanks to this thread I've been playing with Chromaphone a bit, it's really wonderful how it sometimes reacts just as you'd expect it to. There's also a bunch of snare drum presets in there, which are quite tweakable with respect to tone, tonality, timbre, softness etc.

 

It's not really physical modeling, but Image Line's Ogun is great if you're looking for metallic sounds.

[LimpyLoo]

Alright, so I love EZdrummer and BFD and all that sample-based stuff, and they're really great for acoustic drums, but do any decent physical modelled drum VSTs exist? I'm talking realistic stuff here, I know it's probably not easy but I figured there must be something out there, right? It'd be nice to be able to fine-tune the parameters of a simulated acoustic snare, for example.

 

For some reason I'm really really curious why you're specifically interested in physical modeling of drums.

[zkom]

Well, I want to physically model a 20m wide bass drum made of steel and concrete. Samples are bit too hard to find.

[awepittance]

you should think something like that would exist by now but as far as i know, no physical modeling synth has a physics engine powerful enough to customize an objects size and material to be that ridiculous. I wish though, it's always been my dream to have the sound equivalent of Maya. An animated sound rendering engine using real physics and acoustic simulations.

I've always wanted to physical model the sound of a silverback Gorilla slipping on a skateboard at the top of a flight of stairs falling into an upside-down lawnmower at the bottom

[LimpyLoo]

you should think something like that would exist by now but as far as i know, no physical modeling synth has a physics engine powerful enough to customize an objects size and material to be that ridiculous. I wish though, it's always been my dream to have the sound equivalent of Maya. An animated sound rendering engine using real physics and acoustic simulations.

 

I've always wanted to physical model the sound of a silverback Gorilla slipping on a skateboard at the top of a flight of stairs falling into an upside-down lawnmower at the bottom

 

I don't think it's the case that the technology doesn't exist yet, I just think it's not practical for commercial release.

 

Just judging by the state of the art of commercially-available IR (impulse response) plugins (I don't mean reverb plugins but rather the Nebula-type programs that model random esoteric stuff) I would say the technology very obviously exists but CPU issues preclude the more crazy ambitious applications.

[zkom]

I bet you could do it with some of the CAD programs that can simulate vibrations, but not in real time and the SW will cost an arm and a leg

[th555]

I think for stuff like that you should look at science/engineering solutions, rather than commercial audio plugins. Time for some googling.

[futureimage]

I've always wanted to physical model the sound of a silverback Gorilla slipping on a skateboard at the top of a flight of stairs falling into an upside-down lawnmower at the bottom

Might have to attempt this for my final year technical project at uni.

[awepittance]

 

you should think something like that would exist by now but as far as i know, no physical modeling synth has a physics engine powerful enough to customize an objects size and material to be that ridiculous. I wish though, it's always been my dream to have the sound equivalent of Maya. An animated sound rendering engine using real physics and acoustic simulations.

 

I've always wanted to physical model the sound of a silverback Gorilla slipping on a skateboard at the top of a flight of stairs falling into an upside-down lawnmower at the bottom

 

I don't think it's the case that the technology doesn't exist yet, I just think it's not practical for commercial release.

 

Just judging by the state of the art of commercially-available IR (impulse response) plugins (I don't mean reverb plugins but rather the Nebula-type programs that model random esoteric stuff) I would say the technology very obviously exists but CPU issues preclude the more crazy ambitious applications.

 

well show me something which does anything near what i'm describing and i would agree with you. Physical modeling besides in a hand full of specialized circles (mit music department, stanford) is pretty much an exploited technology at this point in time. Technically speaking CPUS and computers are plenty powerful to already do this, so in that sense the technology does exist. It's just the lack of research and development, the software and code doesn't exist.

[ryancolecreate]

drum spillage is also great.

[mcbpete]

At the moment the most advanced physically modelled VSTi I can think of has to be Pianoteq 4.5. A medium leap from 4.0 which itself was a major leap from 3, which was a bloody massive jump from 2, that was pretty darn good already! The amount of controllable parameters in it is simply staggering (oh and yeah it sounds good too)

[LimpyLoo]

 

 

you should think something like that would exist by now but as far as i know, no physical modeling synth has a physics engine powerful enough to customize an objects size and material to be that ridiculous. I wish though, it's always been my dream to have the sound equivalent of Maya. An animated sound rendering engine using real physics and acoustic simulations.

 

I've always wanted to physical model the sound of a silverback Gorilla slipping on a skateboard at the top of a flight of stairs falling into an upside-down lawnmower at the bottom

 

I don't think it's the case that the technology doesn't exist yet, I just think it's not practical for commercial release.

 

Just judging by the state of the art of commercially-available IR (impulse response) plugins (I don't mean reverb plugins but rather the Nebula-type programs that model random esoteric stuff) I would say the technology very obviously exists but CPU issues preclude the more crazy ambitious applications.

 

well show me something which does anything near what i'm describing and i would agree with you. Physical modeling besides in a hand full of specialized circles (mit music department, stanford) is pretty much an exploited technology at this point in time. Technically speaking CPUS and computers are plenty powerful to already do this, so in that sense the technology does exist. It's just the lack of research and development, the software and code doesn't exist.

 

 

 

IR technology enables you to model physical surfaces in kind-of a sonar-esque way. You take some sort of sound and bounce it off a surface and see what it looks like when it comes back. Then you essentially just factor out the original sound and then you can model any sound bouncing off that surface.

 

Also, you can make IR's for synth behaviour (e.g. pitch drift or other idiosyncrasies) or guitar amp behaviour or really anything you could imagine.

 

The technology is pretty commonplace. Sometimes big studios well take an IR of their tracking room to send to another studio that are recording overdubs for a project recorded in the first studio (if that makes sense).

 

So the impracticality probably enters in when it comes to actually rendering the IR that you want to use. You would either have to do it yourself or pay someone else to do it. IMO if you were a decently wealthy you could get it done tomorrow.

[futureimage]

Also, you can make IR's for synth behaviour (e.g. pitch drift or other idiosyncrasies) or guitar amp behaviour or really anything you could imagine.

How do you model pitch drift with IRs? Honestly very interested in this as this is something I'm seriously considering looking into for my tech project - doing some interesting IR/convolution based morphing stuff.

[Guest]

If Iep is still around(?), he could school us on this subject. Holler back...

[LimpyLoo]

 

Also, you can make IR's for synth behaviour (e.g. pitch drift or other idiosyncrasies) or guitar amp behaviour or really anything you could imagine.

How do you model pitch drift with IRs? Honestly very interested in this as this is something I'm seriously considering looking into for my tech project - doing some interesting IR/convolution based morphing stuff.

 

 

Maybe pitch drift is a bad example. (While one could easily model it, it might fall outside the realm of proper IR; but I'm certain you can model the tuning problems of a synth)

 

I think the only requirement is having a constant. So like, you could make an IR of the chorus of a Juno-60 by comparing the chorused signal to the dry signal. And of course the whole point of IR is that it's very thorough and dynamic so one would check how the chorus reacts to low notes and high notes and resonant notes and soft notes and loud notes (and maybe after all you find that the chorus always acts the same).

 

But you can make IR's of specific gear, like tape machines or reverb units or what-have-you, once again seeing how the gear reacts to all sorts of input. I'm pretty sure you could model how an SH-101 behaves on a cold day or how it behaves when you apply pressure to the on/off button.

 

So yes it can definitely be done, but I'm not sure about the method and if it is considered proper IR.

[Guest]

 

So yes it can definitely be done, but I'm not sure about the method and if it is considered proper IR.

 

 

dynamic IR's?

 

Aren't these just a bunch of impulse responses run in succession to catch a certain behavior, for example drifting of tuning?

[LimpyLoo]

 

 

So yes it can definitely be done, but I'm not sure about the method and if it is considered proper IR.

 

 

dynamic IR's?

 

Aren't these just a bunch of impulse responses run in succession to catch a certain behavior, for example drifting of tuning?

 

 

Exactly. I'm just not certain about the terminology.

[awepittance]

 

 

 

you should think something like that would exist by now but as far as i know, no physical modeling synth has a physics engine powerful enough to customize an objects size and material to be that ridiculous. I wish though, it's always been my dream to have the sound equivalent of Maya. An animated sound rendering engine using real physics and acoustic simulations.

 

I've always wanted to physical model the sound of a silverback Gorilla slipping on a skateboard at the top of a flight of stairs falling into an upside-down lawnmower at the bottom

 

I don't think it's the case that the technology doesn't exist yet, I just think it's not practical for commercial release.

 

Just judging by the state of the art of commercially-available IR (impulse response) plugins (I don't mean reverb plugins but rather the Nebula-type programs that model random esoteric stuff) I would say the technology very obviously exists but CPU issues preclude the more crazy ambitious applications.

 

well show me something which does anything near what i'm describing and i would agree with you. Physical modeling besides in a hand full of specialized circles (mit music department, stanford) is pretty much an exploited technology at this point in time. Technically speaking CPUS and computers are plenty powerful to already do this, so in that sense the technology does exist. It's just the lack of research and development, the software and code doesn't exist.

 

 

 

IR technology enables you to model physical surfaces in kind-of a sonar-esque way. You take some sort of sound and bounce it off a surface and see what it looks like when it comes back. Then you essentially just factor out the original sound and then you can model any sound bouncing off that surface.

 

Also, you can make IR's for synth behaviour (e.g. pitch drift or other idiosyncrasies) or guitar amp behaviour or really anything you could imagine.

 

The technology is pretty commonplace. Sometimes big studios well take an IR of their tracking room to send to another studio that are recording overdubs for a project recorded in the first studio (if that makes sense).

 

So the impracticality probably enters in when it comes to actually rendering the IR that you want to use. You would either have to do it yourself or pay someone else to do it. IMO if you were a decently wealthy you could get it done tomorrow.

 

IR's are something totally different than generating a sound from scratch though, they create characteristics of environments to process pre-existing sound from. If you want to look into what i'm talking about more specifically look up waveguide synthesis and the Yamaha VL development.

I've always felt that impulse responses were sort of a dead end, but maybe I haven't heard the examples you're talking about. None of them have sounded as realistic or convincing to me as waveguide synthesis with the proper framework

 

but if I had unlimited funds, as you suggested above I would first hire everybody who worked on the development of the Yamaha Vl1 as a starting point. Recreating that coding in the software realm would only be the first step, but it would be a huge one that still to this day (15 years later) has not been surpassed in quality and realism.

[LimpyLoo]

 

 

 

 

you should think something like that would exist by now but as far as i know, no physical modeling synth has a physics engine powerful enough to customize an objects size and material to be that ridiculous. I wish though, it's always been my dream to have the sound equivalent of Maya. An animated sound rendering engine using real physics and acoustic simulations.

 

I've always wanted to physical model the sound of a silverback Gorilla slipping on a skateboard at the top of a flight of stairs falling into an upside-down lawnmower at the bottom

 

I don't think it's the case that the technology doesn't exist yet, I just think it's not practical for commercial release.

 

Just judging by the state of the art of commercially-available IR (impulse response) plugins (I don't mean reverb plugins but rather the Nebula-type programs that model random esoteric stuff) I would say the technology very obviously exists but CPU issues preclude the more crazy ambitious applications.

 

well show me something which does anything near what i'm describing and i would agree with you. Physical modeling besides in a hand full of specialized circles (mit music department, stanford) is pretty much an exploited technology at this point in time. Technically speaking CPUS and computers are plenty powerful to already do this, so in that sense the technology does exist. It's just the lack of research and development, the software and code doesn't exist.

 

 

 

IR technology enables you to model physical surfaces in kind-of a sonar-esque way. You take some sort of sound and bounce it off a surface and see what it looks like when it comes back. Then you essentially just factor out the original sound and then you can model any sound bouncing off that surface.

 

Also, you can make IR's for synth behaviour (e.g. pitch drift or other idiosyncrasies) or guitar amp behaviour or really anything you could imagine.

 

The technology is pretty commonplace. Sometimes big studios well take an IR of their tracking room to send to another studio that are recording overdubs for a project recorded in the first studio (if that makes sense).

 

So the impracticality probably enters in when it comes to actually rendering the IR that you want to use. You would either have to do it yourself or pay someone else to do it. IMO if you were a decently wealthy you could get it done tomorrow.

 

IR's are something totally different than generating a sound from scratch though, they create characteristics of environments to process pre-existing sound from.

 

In the context of modeling a drum (e.g. a drum made of concrete) then we're just talking about modeling things bouncing off surfaces. So you model a Vic Firth 5b hickory drum stick hitting the dead center of Remo Emporor drum head at such and such a velocity and then bouncing off the concrete and then out into the room and bouncing around a bit and then ending up wherever the microphone or listener is supposed to be. So I would say 'creating a sound from scratch' per se would actually be a matter of modeling a bunch of steps like this.

 

I forget...what specifically were you looking to model?