DIY Filter Question...

Keyboards, Synths, Samplers and Groove Boxes.

Moderators: MetroSonus, Plink Floyd

Re: DIY Filter Question...

Postby gregwar » Fri Oct 11, 2013 10:49 pm

also my toppobrillo dual function generator faceplate came in recently, I still need a pcb/parts

next on the ordering list is a dual low pass gate (I already have a j3rk dual osc and wavefolder)

this all started in a vain attempt to troll don buchla but I've actually learned quite a bit about synthesis and electronics.

I might match those modules up with some 'pre made' ones like make noise dpo, maths, and 4ms pingable env generator (altho there is the option to build that one myself).

I've got a 2 row euro design on modular grid which I can put here if anyone's curious.
User avatar
gregwar
 
Posts: 1726
Joined: Mon Dec 12, 2011 10:59 pm

Re: DIY Filter Question...

Postby ElectricPuppy » Sat Oct 12, 2013 12:24 am

gregwar wrote:you feed an output into an input plus additional components and apparently you can make an osc that tracks pretty good :wat:

http://www.birthofasynth.com/Thomas_Hen ... -4046.html

can sum of u tech savvies have a look-see ?


Looks legit. DO IT
User avatar
ElectricPuppy
The Banned Man
 
Posts: 7099
Joined: Sun May 02, 2010 10:16 pm

Re: DIY Filter Question...

Postby MetroSonus » Tue Nov 05, 2013 5:10 pm

here's a thought.. they make convolution based VST reverbs that can load reflection images, which are a tone subtracted from the reflected sound.

Why cant' they do that with filters? Run a straight tone through a filter, subtract the tone and then use the resulting sound as a basis of the model for a VST base filter?
User avatar
MetroSonus
My style, is the best style.
 
Posts: 8455
Joined: Tue Mar 09, 2010 6:20 pm

Re: DIY Filter Question...

Postby ElectricPuppy » Tue Nov 05, 2013 6:08 pm

I think they can do that with filters, but the analysis is different.

For reverb, I believe the analysis is based around time; it's temporal analysis. They set off a sharp transient (like a balloon popping) and record the result. The resulting wave shows all of the pop's reflections over time.

For filters, I think the analysis is based on frequency, not time. To analyze a filter set a specific cutoff, you would run a sine sweep through it and observe the results. That would show you the filter's response at THAT setting. You would repeat this, dialing down (or up) the filter's cutoff point and get a different result each time. You'd wind up with a 3D mapping of the filter's response over all combinations of input frequencies vs. filter cutoff frequency. Then it should be possible to use that mapping to mimic that filter's response over any arbitrary input.

Disclaimer: I'm not an expert on these things.
User avatar
ElectricPuppy
The Banned Man
 
Posts: 7099
Joined: Sun May 02, 2010 10:16 pm

Re: DIY Filter Question...

Postby selfinflikted » Tue Nov 05, 2013 6:12 pm

ElectricPuppy wrote:Disclaimer: I'm not an expert on these things.


Damn sure fooled me!
User avatar
selfinflikted
Down in it
 
Posts: 4656
Joined: Fri May 25, 2012 11:32 am
Location: Mobile, Alabama

Re: DIY Filter Question...

Postby MetroSonus » Tue Nov 05, 2013 6:18 pm

well yes.. that's exactly what I meant. Filters are going to be different as they have many different potential settings, but the reverbs more or less assume the sound is coming from one place. Most that I know of don't account for things like near the wall or something.

But I guess the question is, why hasn't anyone done it? :mad:
Or maybe that's exactly what they do and they just haven't made the filters I want yet? :confuse:

But regardless, I have yet to hear anything totally convincing. But I do love me some VSTs though.. and yet I hate them.

I don't know.

pepsi, coke, chicken or fish...
User avatar
MetroSonus
My style, is the best style.
 
Posts: 8455
Joined: Tue Mar 09, 2010 6:20 pm

Re: DIY Filter Question...

Postby gregwar » Tue Nov 05, 2013 9:38 pm

nebula has done this with a variety of filters, eqs, consoles, tape sims, etc

it has something to do with their n.a.t. sampler and something called 'vectorial volterra kernels'. the process is similar to convolution reverbs but teh kernels work differently than impulse responses. basically you send a signal through x piece of gear and record it in the n.a.t. sampler but i'm not sure if its a sine sweep, white noise, or some other kind of test tones.

i do know it captures the gear at various volumes and can be more 'dynamic' and can capture and recreate distortion, etc (the kernels apparently capture up to 9 harmonics, etc). i would use these to add 'tone' and 'mojo' to a recording.

there are a bunch of third party libraries here's a sum filter options from signaltonoize:

moog 4 pole: http://rhythminmind.net/STN/?page_id=1733
vintage "l" filter: http://rhythminmind.net/STN/?page_id=2361
t-rez filter: http://rhythminmind.net/STN/?page_id=1728

i've read that the downfall of this technology is time-based effects like modulation don't really work. on the links you can hear demos of stn going through filter sweeps and whatnot. i think it sounds pretty convincing :idk:

nebula has a built in lfo and some 3rd party programs like cupwise stuff makes great use of it with bandpass and notch filters, tremolos, and a tube radio low pass :confuse:
http://www.cupwise.com/cup/
User avatar
gregwar
 
Posts: 1726
Joined: Mon Dec 12, 2011 10:59 pm

Re: DIY Filter Question...

Postby soundwave106 » Wed Nov 06, 2013 4:12 pm

MetroSonus wrote:Or maybe that's exactly what they do and they just haven't made the filters I want yet? :confuse:


Most analog electronic filters in synthesizers, I believe, are infinite impulse response type filters. That means that, although you *could* model a portion of the impulse response, you would never be able to 100% capture the entire thing.

Although I'm sure there's probably a practical limit that you could record and simulate even with an IIR filter, the impulse would probably have to be both very large and sampled over a wide range of tone frequencies, resonance settings, and harmonic content (3 dimensions). Again, you probably wouldn't need to capture *everything*, maybe key frequencies + clever interpolation is enough, but I would think a fair bit would be necessary for a decently accurate model. This also would imply, I think, that the harmonic content and frequency of the input would have to be analyzed realtime to determine the final output (many simplifications could be made I suppose but still...)

This would be more complicated, I think, than convolution impulse reverbs. I think it's possible, yes, but it sounds heavy on the CPU. The model could probably be simplified greatly, but I guess that's why the trend du jour for more accurate filters is the SPICE approach (emulate the analog circuit at the component level).

Even then, analog filters will always be "faster" than digital filters in one sense, as digital filters always have inherent latency in their calculations. (Analog filters may be "imperfect" and are less idea for extremely precise tone shaping, but they are fast! Plus many people are pleased by the "imperfections" of analog.) An impulse won't resolve that issue.

Disclaimer: I'm no major DSP math wiz expert here, so it may be simpler than I think... it just doesn't sound that simple at all at first glance, at least for something usable realtime...
BC | SC
User avatar
soundwave106
 
Posts: 667
Joined: Mon Oct 15, 2012 1:09 pm

Re: DIY Filter Question...

Postby Veracohr » Wed Nov 13, 2013 12:43 am

Isn't that what analog modeling is all about? Modeling (perfectly) the response of an analog filter would require a theoretically infinite number of models at different settings, but the properties of analog components are understood well enough that a mathematical model of the response can be calculated rather than observed. Something simple like a second order low pass filter can be described by

Image

But that's a first-order model. It can be made much more detailed by including imperfect properties of the active components, parasitic effects, etc. And all that can be done mathematically. After all, isn't that how Universal Audio has been able to so faithfully recreate analog hardware?
User avatar
Veracohr
 
Posts: 181
Joined: Fri Jan 11, 2013 7:55 pm

Re: DIY Filter Question...

Postby Plink Floyd » Wed Nov 13, 2013 6:31 am

Why are stand alone hardware filters so expensive? Just low demand? Is the $300 MoogerFooger the most affordable? Not everyone has a voltage controller gizmo, so why isn't there a midi controlled one? Where is the $50 Korg Kilter™?

Dammit Spock I want answers!
User avatar
Plink Floyd
Moderator
 
Posts: 6318
Joined: Wed Apr 28, 2010 6:13 am

Re: DIY Filter Question...

Postby selfinflikted » Wed Nov 13, 2013 2:02 pm

I firmly object to this many maths in this thread.
User avatar
selfinflikted
Down in it
 
Posts: 4656
Joined: Fri May 25, 2012 11:32 am
Location: Mobile, Alabama

Re: DIY Filter Question...

Postby MetroSonus » Wed Nov 13, 2013 2:33 pm

Why are stand alone hardware filters so expensive?


That's cause it's Moog. The Apple of synths...

Get whatever flavor filter you want in Eurorack, including Moog. The average price for a Euro module is 200$. Not counting the 200$ roughly for a power supply, rails ect.
User avatar
MetroSonus
My style, is the best style.
 
Posts: 8455
Joined: Tue Mar 09, 2010 6:20 pm

Re: DIY Filter Question...

Postby soundwave106 » Wed Nov 13, 2013 4:30 pm

Veracohr wrote: but the properties of analog components are understood well enough that a mathematical model of the response can be calculated rather than observed. Something simple like a second order low pass filter can be described by
Image


Yes, that is the point of modeling indeed.

The problem with the perfect theoretical equations as I understand it are that they don't capture some of the "nuances" of the analog filter (usually errors that the analog circuit design introduces).

Of course, as you say, emulating the circuit components *will* capture those "nuances". Which is why, as you said, the trend du jour is SPICE-like simulations where the actual circuit design is emulated vs. throw the signal into an equation and be done with it.

There still is a problem with even a Spice simulation though: the latency of digital filter design. This is why Diva and others make a big deal about the "zero delay feedback", the "feedback" in this case probably means resonance (maybe some other areas too). In many cases in analog circuits, resonance is a phase-reversed feedback loop in the circuit. A simple digital feedback loop will always calculate on the previous sample, *not* the real-time sample, which introduces "problems".

(In fact, *any* sort of digital calculation that relies on loop will have this problem. Another well known trick in modulars / analog patching synths is to feed the post-filter output into the mixer. In some analog synths, this makes for a really nice growly sound. I don't think I've seen a digital emulation that can handle this well, yet, and I guess this is in part because of that aforementioned problem.)

There are of course probably ways around them, as the several 0d filters in VSTs have shown.

Even with these difficulties, I think it would be much easier to model vs. compute a huge impulse...
BC | SC
User avatar
soundwave106
 
Posts: 667
Joined: Mon Oct 15, 2012 1:09 pm

Re: DIY Filter Question...

Postby MetroSonus » Wed Nov 13, 2013 6:03 pm

This is just an arm chair opinion so take it for what it's worth...

Software, for whatever reason, doesn't capture the breadth of sound in any sort of hardware gear to me, VA or RA. And I think this is where the point of software is. It sits in the mix, there's not much you have to do to it and that's the way most people want it, I think. Easy, cheap and not a lot to think about.

Can you do that with external gear? Sure. But I guess us synth heads are in the minority and we like the option to go above and beyond that. I personally would rather write music (or create sounds) around the nuances you hear in audio than sit down and do a cook book patch. That's why I like synths.

The other thing too and this goes with what you were saying SW106, is that the circuits age. A majority of synths never really sounded "fat" until they started getting old.. When they were new, they were what they were supposed to be, precisely tuned electronic insruments; and I mean that more in a scientific way than a musical sense.

I've seen enough of it to prove my point, to an extent. I've had so many poly 800's over the years. One in particular made made such a gut busting bass with a huge "gwaaaaaaaaaaaaaH' sound when you swept the filter. Then that one broke and I got rid of it because they were cheap enough on ebay.. And I never found one that made the same sound again. And then I encountered EPs prime specimen of an EX800 with absolutely no noise to the outputs or anything and really, f*ed all if I could hear the difference between that and synapse audio's Poly 850 plugin.

There are a few standouts that make *that* sound that makes you lust after analog. But for a finely tuned machine in a polished and produced song, you wouldn't know the difference.
User avatar
MetroSonus
My style, is the best style.
 
Posts: 8455
Joined: Tue Mar 09, 2010 6:20 pm

Re: DIY Filter Question...

Postby soundwave106 » Thu Nov 14, 2013 2:15 pm

Yeah, I don't think that most models include a "capacitors age and knobs get weird etc. mode". :) When that happens, sometimes the results are undesirable, but sometimes there's "something there".

Most of what I write has a healthy mix of analog and digital. Digital has the benefit of being able to use stored samples, after all, and makes some terrific sounds. I guess the only oddball thing for my setup is that most of my digital lives in the PC (eg Omnisphere, Alchemy, etc.), except for some digital oscillators within the Eurorack (the modular is more fun to play with than compose with though.)
BC | SC
User avatar
soundwave106
 
Posts: 667
Joined: Mon Oct 15, 2012 1:09 pm

Re: DIY Filter Question...

Postby Veracohr » Fri Nov 15, 2013 8:25 pm

soundwave106 wrote:The problem with the perfect theoretical equations as I understand it are that they don't capture some of the "nuances" of the analog filter (usually errors that the analog circuit design introduces).


I guess that was my point. That if someone cared enough to put in the time and effort, they could model all those component irregularities and imperfections that contribute to the actual (not theoretical) filter response, and construct a (probably very complicated) equation that could potentially be much closer to the sound of an analog filter.
User avatar
Veracohr
 
Posts: 181
Joined: Fri Jan 11, 2013 7:55 pm

Re: DIY Filter Question...

Postby gregwar » Fri Nov 15, 2013 9:23 pm

just turn up the analog knob on ur plug-in duh :thumbs up;
User avatar
gregwar
 
Posts: 1726
Joined: Mon Dec 12, 2011 10:59 pm

Re: DIY Filter Question...

Postby selfinflikted » Fri Nov 15, 2013 9:29 pm

Hey Veracohr , weren't you working on some kind of effects box for a class? How's that coming along?
User avatar
selfinflikted
Down in it
 
Posts: 4656
Joined: Fri May 25, 2012 11:32 am
Location: Mobile, Alabama

Re: DIY Filter Question...

Postby Veracohr » Thu Nov 21, 2013 6:26 pm

selfinflikted wrote:Hey Veracohr , weren't you working on some kind of effects box for a class? How's that coming along?


Yes, I'm eyeball deep in the design stage right now. It's my senior project, so it's a 3-term process and I should have the final working unit in the spring (at the latest). I'm a little miffed by the schedule I have to work within, because it requires the design to be completed in just a few weeks, so I'm busting my balls right now even though there will be two whole terms where I just have to construct and test it. I suppose the schedule might work better for different types of projects, but I've had 4 weeks to research and design 4 different circuits that meet my self-imposed requirements while accounting for the limitations of the OTA chips I'm using, choose specific parts where it matters (like the microcontroller, DACs, etc.), prepare complete schematics and update my proposal. While also putting time into my other class and working full time.

I'm a great proponent of complaining. It makes me feel better.
User avatar
Veracohr
 
Posts: 181
Joined: Fri Jan 11, 2013 7:55 pm

Re: DIY Filter Question...

Postby selfinflikted » Thu Nov 21, 2013 6:45 pm

Veracohr wrote:I'm a great proponent of complaining. It makes me feel better.


Indeed. I am the same. Some times I complain just to hear myself talk. I, for one, will be anxiously awaiting your results!
User avatar
selfinflikted
Down in it
 
Posts: 4656
Joined: Fri May 25, 2012 11:32 am
Location: Mobile, Alabama

PreviousNext

Return to KSS

Who is online

Users browsing this forum: No registered users and 1 guest

cron