Filters apt for Physical Modeling?

[Ghost_the_garden]

Hey, so this is more of a "Help me pick a filter thread" than a quest for the best pm filter. I'm looking to make Haunting woodwind sounds, deep cellos, bagpipes, and banshees. I'm not really interested in squelchy resonant sounds at all, or vintage wankery. The E340 is my only oscillator atm so thats what i'll be filtering, and for awhile this will be my only filter. Really looking for something that sounds alive, and unpasteurized, or could otherwise work with the cloud generator to achieve these characteristics.

These filters caught my attention though i'm certainly not sold on anything

Doepfer 127

Doepfer 104 - no cv!

Res 4 - Too expensive to realistically consider actually.. :-<

Three sisters - This one sounds real beautiful when self oscillating, but I need a filter! lol Otherwise it seems like one of the more interesting filters in euro, its just most the demos didnt sound as organic as I like, but maybe it can be patched further in that direction?

R-54/52 - Not sure I want to spend this much, but they do sound very alive!

[Footkerchief]

To be honest, I'd look more at wavefolders and waveshapers for those timbres. I get great results from my Ultrafold, Barton Dual Rectifier, and FM AID. Not sure how they'd interact with the E340, but worth exploring.

[damase]

Mutable Rings is really good at the kind of sound you describe if you use it kind of like a filter with the audio input and maybe a vca. and thats just one of many easy ways to get awesome sounds from it

ifm sprott is very alive and experimental... has a distinct color, but very organic. comes with a nice vca on the input.

[cbeefheartuk]

I haven't used one yet but maybe the random source - Serge resonant eq might be a good choice for addng an acoustic feel to sounds and with the feedback you should get some really interesting tones.. i think i will get one soon and i am wanting to make sounds like you describe with it hopefully. DIY only though i think.

[MindMachine]

EMW Multi Bandpass VCF is a low priced odd ball that can be used for those types of sounds. Resonance is fixed for all three filters, but they each have VC of the frequency. I sometimes use an EMW 3 Band Splitter into the Triple Band pass via mixers/VCA's:

[listentoaheartbeat]

For those kind of sounds, I'd rather use noise + tuned delay with a filter in the feedback path. Subtractive synthesis is not physical modeling. You can approximate acoustic instruments with a subtractive synth, but I find the level of abstraction to be rather high. Look at FM, wave folding, and tuned delay applications for more plausible 'acoustic' sounds.

[sihiL]

To my ears, LPGs sound much more natural than most filters, especially for percussive sounds.

[timoka]

^this. i would look into wavefolders first, i don't have a cloud generator but in my imagination it must sound great with sine waves spread just a bit and then into the twf or another folder. or the r-54 which is not only a filter but also a waveshaper and oscillator and more alive than anything i know.
but like listentoaheartbeat wrote i would look more into the beginning of the path and not use an oscillator but an exciter and then tuned delays.
mungo c0 can load any filter shape you like, if you know logic's sculpture synth, then you know that the body eq setting on that synth can change the sound drastically and the c0 can do that for you, so i would go for exciter - delays - filtering for acoustic sounds...

[timoka]

here is something i am experimenting with for the last days, this above mentioned body eq thing with the c0:

http://soundcloud.com/uzala/c0-clarinet-eq-test

square wave into the c0 loaded with an IR from a matched clarinet eq.
played with a ribbon controller. at the end the same IR but distorted and a narrow pulse wave into the c0...sounds like a brass instrument.

[geecen]

Sounds like you need a Fonitronik PS3100

Triple resonant band pass with vactrol control and built in lfo. They're DIY mainly, but you could get someone to build you one on the DIY board or try and buy one second hand.

[kirklandish]

i don't have a cloud generator but in my imagination it must sound great with sine waves spread just a bit and then into the twf or another folder

I can confirm that it does sound amazing (I have the uFold v2), though not always the most "natural." This combo often leads me into more My Bloody Valentine wall of sound tones (maybe that's just me though).

Also agree that LPGs sound more natural than filters.

[trip]

Mutable Rings and Elements both do a kind of physical modelling. Rings is just the filterbanks,you provide the noise source / sound to be filtered, elements is the filterbanks plus three diff types of complex noise source. Both sound amazing, and I'm pretty sure they're the only modules which do that kind of thing in euro - would need waaayy too many filters and precise modulation sources for any kind of 'analogue' physical modelling, which isn't to say you can't create some fairly convincing analogues to actual instruments with vcos and vcfs. Check out some demos of elements, I'm pretty sure that would be the most effective module to do what you want, there's an endless amount of sounds you can get out of that thing - PLUS an awesome fm voice in the easter egg mode, can't go wrong.

[trip]

ps. you're totally on the right track with three sisters as well, physical modelling is generally a bunch of band pass filters interacting with one another, which is what three sisters kind of does on a small scale. Rob Hordijk's twin peak is low pass or two resonant band pass filters and anywhere in between. Fucking great for all kinds of woody / throaty / water drop stuff, great for 'natural' sounding things, not so great for screaming resonant analogue stuff. Ian Fritz threeler is crazy good for multiphonics / chaotic stuff too, again it's three filters in a variable topology.

[thermionicjunky]

Most frequency-domain physical modeling is done digitally with a large number of filters. A resonant filter bank such as the Serge Resonant EQ could be useful. If Random*Source ends up releasing a Euro version of the Haible String Filter, I would recommend that. It goes well with folders, LPGs and variable filters. It makes everything sound wooden.

For Karplus-Strong based modeling, check out Ron Berry's patches. They require a good flanger, but otherwise work well with standard modules.

My favorite physical models are done in the time-domain. This includes Madrona Labs' Kaivo plugin and Ian Fritz's PM clarinet circuit. The PM clarinet uses standard VCO and VCF circuits combined with some unusual custom circuits for the bore (sample/delay/convolve) and reed/mouthpiece. I love time-domain modeling because it so easily leads to chaos (nonlinear feedback loop).

[timoka]

kaivo is wonderful! never heard of the pm clarinet circuit...do you have a link to some additional infos? thanks!!

[thermionicjunky]

kaivo is wonderful! never heard of the pm clarinet circuit...do you have a link to some additional infos? thanks!!

It used to be up on Ian's website. I archived the schematics about 5 years ago and finally built it in 2015-16. Ian may share the schematics with you if you ask him nicely.

It requires a saw-core VCO (upward ramp) which determines the sample-rate of the bore circuit (and the fundamental frequency). The bore is essentially a complex BBD made from analog switches and S&H circuits. The bore output is smoothed with a v/octave lowpass filter and fed back through special analog computation circuits that model the interaction of mouth pressure and wave pressure and the nonlinear transmission function of the reed. The final output is then typically passed through another LPF to roll off some high frequencies.

I modified the design a bit to add patch points and full voltage-control. Instead of 2/4-pole LPFs made with the CA3280, I used 3-pole LPFs made with the SSM2164. I added 2164 VCAs and crossfaders to automate all manual controls and I took Ian's variable sync circuit from the Teezer and added it to his mosfet saw-core VCO.

The circuit can faithfully emulate a clarinet. Just like a clarinet, it sounds most clarinet-like at lower frequencies. Even though the output looks just like a rounded square wave on the scope, the way it responds to mouth pressure is very realistic. With a bit of adjustment, patching, and/or modulation it goes way beyond a clarinet emulation.

[tbecker]

The res 4 is a really good filter for pm. In addition to the standard ping into the input, you can do am and fm on each band and with careful selection of the bands that are being used via the individual level knobs, you can use the notch output to fm a band while using the band output. This is sort of like pingable amplitude with a pingable fm. These techniques lead to excellent bass drums, snares, toms, blocks, marimbas, and general percusions. I think the more compex resonant bodies like a clarinet, or cello are difficult with even several resonant bandpass filters. The Mungo C0 seems to be designed for this purpose and is less than a res 4 but still more than most standard filters like the three sisters. Good luck with what ever filter you select, I think you can never have enough!

[timoka]

i just realized that my example of the c0 was unbelievable bad haha, here is a new one with a clarinet eq loaded:

http://soundcloud.com/uzala/c0-clarinet-eq-test

square wave played with a ribbon contr. into the c0. at the end i distort the c0 with it's built in amp and feed it a narrower pulse wave...more like a brass instrument with a mute now.

[sihiL]

i just realized that my example of the c0 was unbelievable bad haha, here is a new one with a clarinet eq loaded:

http://soundcloud.com/uzala/c0-clarinet-eq-test

square wave played with a ribbon contr. into the c0. at the end i distord the c0 with it´s built in amp and feed it a narrower pulse wave...more like brass now.

Impressive! Is it an impulse response or some type of EQ-match thing? I'm unfamiliar with the c0. What did you feed it besides the square wave?

[timoka]

it is an impulse response file i made from a matching eq of a clarinet which plays two octaves i think.
there are quite a few good anechoic soundfiles of instruments available and i just made this matched eq
from one of them. after that i played a pure impulse through this eq and record that result,
cut and export and you have your IR. the good thing about this is it takes up almost no sample space at all,
so you can create your IR in 96khz or even higher and then downsample/pitch on the c0 for different timbres of the same instrument...all in real time!

nothing but the square wave into the c0.

[timoka]

kaivo is wonderful! never heard of the pm clarinet circuit...do you have a link to some additional infos? thanks!!

It used to be up on Ian's website. I archived the schematics about 5 years ago and finally built it in 2015-16. Ian may share the schematics with you if you ask him nicely.

It requires a saw-core VCO (upward ramp) which determines the sample-rate of the bore circuit (and the fundamental frequency). The bore is essentially a complex BBD made from analog switches and S&H circuits. The bore output is smoothed with a v/octave lowpass filter and fed back through special analog computation circuits that model the interaction of mouth pressure and wave pressure and the nonlinear transmission function of the reed. The final output is then typically passed through another LPF to roll off some high frequencies.

I modified the design a bit to add patch points and full voltage-control. Instead of 2/4-pole LPFs made with the CA3280, I used 3-pole LPFs made with the SSM2164. I added 2164 VCAs and crossfaders to automate all manual controls and I took Ian's variable sync circuit from the Teezer and added it to his mosfet saw-core VCO.

The circuit can faithfully emulate a clarinet. Just like a clarinet, it sounds most clarinet-like at lower frequencies. Even though the output looks just like a rounded square wave on the scope, the way it responds to mouth pressure is very realistic. With a bit of adjustment, patching, and/or modulation it goes way beyond a clarinet emulation.

thanks, your description sounds like something i want! is it possible with it to have something like an embouchure delay...if you put too much pressure in it it overblows...changes the fundamental frequency?
anyway, i think i will write mr fritz!

[timoka]

these tutorials are very helpful for physical modeling, even though most stuff isn't really possible within modular it is still a great way to find new and interesting patches, if you know reaktor, then you might know the author from his astounding physical modeling instruments:

http://electro-music.com/pm_tutorial/Index.htm

[thermionicjunky]

thanks, your description sounds like something i want! is it possible with it to have something like an embouchure delay...if you put too much pressure in it it overblows...changes the fundamental frequency?
anyway, i think i will write mr fritz!

Extreme overblowing via the pressure parameter forces the reed aperture to close just like an acoustic clarinet. I haven't noticed changes to the fundamental, since that is mostly governed by the pilot VCO and the feedback filter. Also, you can't really get typical clarinet multiphonics without the key system, but there are numerous ways to get complex results through overdrive, feedback, FM, sync, and "misadjustment" of the various parameters.

[vytis]

Here's some plucky stringed instrument I've made out of 9U of Doepfer modules:

[soundcloud url="https://api.soundcloud.com/tracks/277485870" params="auto_play=false&hide_related=false&show_comments=true&show_user=true&show_reposts=false&visual=true" width="100%" height="450" iframe="true" /]

It would be great to follow this up with A-127/A-104, timoka's suggested c0 or an impulse response reverb for some increased realism.

In my opinion the beauty of physical modelling is not in how similar it is to reality, but in how different it is from our expectations and the uncanny feeling this creates.

Timoka - your c0 examples are pure beauty (like the rest of your music)! Unsettling.

[timoka]

thanks vytis,
that's a lovely example! astounding how it sounds like karplus strong even though it isn't, very very good example to show where fm and wavefolding can take you!!
and i agree with you, i really love physical modeling where you are able to make impossible instruments, a mile long flute, a concret trumpet, a tiny plastic violin etc.
that's why i still need and love my computer with kaivo, reaktor and aalto on it :-)