Physical Modelling from a circuit schematic

[AdmiralQuality]

Here... I posted this a while back for the Music Cafe board's "First Song" thread ( http://www.kvraudio.com/forum/viewtopic.php?t=145241 ).

http://www.admiralquality.com/mp3/AQs2n ... mp3?ampsim

This is me, in 1985, playing "guitar" on synthesizer -- a Roland JX-3P. The "amp" distortion is transistor distortion, from a crappy old Traynor PA. There's also spring reverb from the PA. And there's some tape saturation too. The drums are a Roland TR-606 (also intentionally under heavy distortion from the PA). Recorded on a Fostex X-15 4 track cassette (that I got for X-mas the day before this was recorded! )

Oh... there's also a guitar power chord type sound in this other one, the first song I recorded on the 4 track, Dec 25, 1985. http://www.admiralquality.com/mp3/AQsFi ... mp3?ampsim It's the same gear. Roland JX-3P, possibly with some added distortion from overdriving the PA's inputs, though that "metal guitar" patch I made on the JX-3P sounded pretty distorted anyway -- was pretty much only good for playing power chord 5ths and chuggy stuff.

[DavenH]

Synthesis is my goal certainly, but, if using physical modelling, the difference between emulating an amplifier and emulating the guitar and amplifier (guitar through the amp) is almost trivial. So I'm after an amp emulation, then a clean guitar emulation. I'm hoping the latter won't be much of a problem.

WOAH! You've got that TOTALLY backwards. Amp emulation is what's trivial (or at least the easier part). Synthesis of convincing guitar tones (and even worse, a way to CONTROL it!) is like the holy grail. AFAIC nobody's even come close yet.. though many have tried.

That's quite.. no, really confusing. A clean guitar tone is so simple compared to the distorted tone, both in harmonic spectrum and waveform. Moreover, there's no need for convincing guitar tones - they are going to be distorted beyond all recognition. I do have some context here, I'm not just making blind assumptions.

Let me give you an example...

Here's are two clips:

Clean tone
Distorted tone

The first is a few sawwaves through a filter - time to synthesize: about 5 minutes for proper adjustments.

The second is the same sawwaves through a 27 unit FX chain (in FL Studio) - experimentation with filters, eq, dist plugins, compression, routing - 3 years and ongoing.

So you can understand why I am more concerned about the amp emulation!

I thought you just wanted an effect to process a real guitar through! Really, the sound doesn't come from the amp, it comes from the guitar.

With reference to the preceding example, I'll respectfully disagree - I can feed all sorts of things through that FX chain, and it sounds almost the same. I just need something that's close enough to a guitar to sound right when amplified and distorted.

Heh, yep. Even worse, more instruments crowded around the guitar playing a chord. I don't have a mesa boogie myself to test with, just clips from songs. I'm pretty good at pattern recognition, but there are altogether too many variables for which to account when analysing such a complex sound in the midst of overlapping elements. Now if I had the monophonic, clean guitar tone and the distorted tone on their own, it'd be easier to derive the function to convert the two.

Oh, then you're screwed. Looking at recorded music on an oscilliscope or FFT won't do a THING for you. Your EARS are MUCH better at pattern recognition than your eyes.

Well, I agree! That's why I need to investigate physical modelling instead, which is why I came up with the thread

[AdmiralQuality]

Well neither of those sound guitar like. But it would sound a lot MORE guitar like if you started with a guitar sound instead of a very obvious synth sound.

Sorry, you're just wrong. You'll learn...

I mean really, why not just buy a copy of Guitar Rig? And buy a guitar? Your premise that it's possible to take that synthy sawtooth sound and somehow guitar-ify it is why it's taking you 3 years to make a bad sound. Start with a guitar sound, and you can distort it with any number of products, and it'll sound like a guitar through an amp.

There's no such thing as "physical modelling" by the way. (Well, not unless you're an aeronautical engineer or something.) It's marketing hype. Yes there are products that claim to use it, and there's some stuff that's KIND of physical modelling... but again it's so simplified (and so focused on the RESULT, rather than reproducing the CAUSE) that it's not really a "physical" model of anything.

[DavenH]

Regardless of sound, it should have proved my point. There's no need to argue more about that.

I do have a guitar - a Gibson actually, but guitar rig is too expensive for me at the moment. Even if I did have the funds, it doesn't really do what I want, as far as I can tell from the demos.

About physical modelling... huh? Where'd you get that assertion? Modelling the physical properties of an instrument has nothing to do with physics? Simplification doesn't make it not a model.

[AdmiralQuality]

Nothing to do with physics. Everything to do with math.

And to your point. You seem to think real gear would do what your 27 unit FX chain does. Which was why you wanted to physically model it. If you ran that synth sound through a real Boogie, it wouldn't come out sounding like your demo. (Furthermore, your demo sounds like someone's playing fifths. Using a pitch shifter in there somewhere?)

So, real guitar amps make real guitars sound distorted. Run your Gibson through your 27 FX chain and what does it sound like? Probably not so good. (Or maybe it does? But again I think you've got a harmonizer/pitch shifter in there somewhere that would futz up any chord you attempted to play.) So whatever you mean by the term, why use "physical modelling" to do something so unreal? I think you're into the realm of metaphysial modelling... or something.

Anyway, as long as it makes you happy. S'what it's all about. Cheers.

[AdmiralQuality]

And just to more properly define the term. "physical modelling" is when you model something *physical*. Something made of matter, with moving parts. And run essentially a piecewise physics simulation on those parts. Here's one of the few products I know of that actually uses something CLOSE to physical modelling http://www.applied-acoustics.com/stringstudio.htm Arturia's Brass might too, I don't know much about it. But still, in either of those products, they're not modelling air molecules, or even fluid dynamics, or pieces of an object (except for a VERY simplified string model.)

http://en.wikipedia.org/wiki/Physical_modelling <-- not applicable to electronic instruments

But "physical modeling" isn't usually what you'd call modelling an electronic circuit. Like, what, am I supposed to code the number of electrons stored in a capacitor? The shape of the magnetic field around a transformer? The heat dissipated by a resistor? No... because the PUROPOSE of those circuits distills down to much simpler math than the math required to describe the PHYSICS of what's happening. The analog circuit is just a very clunky way to do math, with the "fluid" of electricity. In computers, you can just do it with pure information. So a clean amp is just a single multiplication op... I don't have to model transistors and resistors and power supplies and impedances... But it's exactly the FLAWS of analog electronics that most of us miss in the digital realm. So that's why they make amp sims. I'd suggest you invest in one. The work is already done. But again, maybe you don't like them because they model, er, emulate REAL amps... and real amps don't do what you seem to want. I used to play SH-101 through a Fender Twin Reverb... and it didn't make my synth solos sound guitar like AT ALL! Not even twangy! (Though a Twin does make a GREAT amp for an analog synth!!! )

So, how that distorted synth sound proves any point, I dunno. Guitar amps don't make synths sound like guitars. They're not guitar-ifiers. They're just amps intentionally designed to make pleasant sounding distortion. So how is modelling, physically or otherwise, the properties of a device that doesn't do what you want going to help you?

But again, as long as you're having fun.

[DavenH]

Nothing to do with physics. Everything to do with math.

We must be talking about different things... I'm not talking about the demo here, but stuff like sound on sound's synth secrets, don't you consider that physical modeling? One of their articles on pan pipes

And to your point. You seem to think real gear would do what your 27 unit FX chain does. Which was why you wanted to physically model it. If you ran that synth sound through a real Boogie, it wouldn't come out sounding like your demo. (Furthermore, your demo sounds like someone's playing fifths. Using a pitch shifter in there somewhere?)

So you're sure that nothing but a real-sounding guitar would produce the correct output from the Mesa? I guess I could live with that, since I do have a real guitar, but it would still be nice to keep it all digital.

There's no pitch-shifting in the demo... it's taking the synth input just as it is. That said, it doesn't take chords very well, they become too noisy. Of course, I'm not satisfied with it (it does sound pretty crappy compared to a lot of tones), else I wouldn't bother trying to emulate a Mesa

So, real guitar amps make real guitars sound distorted. Run your Gibson through your 27 FX chain and what does it sound like? Probably not so good. (Or maybe it does? But again I think you've got a harmonizer/pitch shifter in there somewhere that would futz up any chord you attempted to play.)

I can't because I don't have recording equipment, but I checked with a couple of guitar samples, and everything sounds pretty much the same. I think that's because I have it filter off most harmonic content before distorting, otherwise you get really harsh, unpleasant sounds. If amplifiers don't have prefilters, then I guess you're completely right... I assumed they did.

Anyway, as long as it makes you happy. S'what it's all about. Cheers.

True

I never said I was happy with it though, I'm not. Just a demonstration...

[VladimirDimitrievich]

The second is the same sawwaves through a 27 unit FX chain (in FL Studio) - experimentation with filters, eq, dist plugins, compression, routing - 3 years and ongoing.

the name is Daven ... uses FL Studio ... huge FX Chain ... simulates guitar stuff ... are you the Daven whos demo song is included with FL6?

[DavenH]

The second is the same sawwaves through a 27 unit FX chain (in FL Studio) - experimentation with filters, eq, dist plugins, compression, routing - 3 years and ongoing.

the name is Daven ... uses FL Studio ... huge FX Chain ... simulates guitar stuff ... are you the Daven whos demo song is included with FL6?

Yes.. and thanks, but what's immensely frustrating is that the included song isn't the correct version... if I'm not mistaken, there's a missing sample, and all the Sytruses in that song load the default preset! (I was testing the beta at the time the installer was put together)

[AdmiralQuality]

Nothing to do with physics. Everything to do with math.

We must be talking about different things... I'm not talking about the demo here, but stuff like sound on sound's synth secrets, don't you consider that physical modeling? One of their articles on pan pipes

That SOS article is just all demonstrated with classic subtrative synthesis. Physical modelling would be if you had some algorithm that modelled air waves in a tube. You'd have controls for tube length, material, number of holes... breath angle... you know, PHYSICAL stuff. But envelopes and filters and qauare wave oscillators... nothing physical modelling about that!

Trying to reproduce a real world sound is an APPLICATION. You can attempt to do that with ANY form of synthesizer. Physical Modelling is a TECHNIQUE, and defines the type of synthesizer. (And an overhyped technique). A synthesizer can be Physical Modelling, just like it can be Additive, Subtractive, FM, or whatever other technique they choose to design it around. But even the ones that claim they ARE physically modeled, aren't, really.

S'all.

And I'm convinced that someday this technique WILL be real, when we have power to spare. But that's assuming the computers aren't so smart by then that they've already exterminated us humans.

[AdmiralQuality]

I can't because I don't have recording equipment, but I checked with a couple of guitar samples, and everything sounds pretty much the same. I think that's because I have it filter off most harmonic content before distorting, otherwise you get really harsh, unpleasant sounds. If amplifiers don't have prefilters, then I guess you're completely right... I assumed they did.

Oh, and amplifiers do have "pre-filters". It's called the tone stack. (You know, Bass, Middle, and Treble knobs. That go to 11.) But still, you don't usually have to roll off ALL your highs even if going for crunchy thrash? slash? metal tones.

And the problem with heavy EQing like that, is the lower pitched strings will sound brighter relative to the high ones (which might be filtered right out of existence... cuz the tone filters have no way of tracking the incoming pitches.) So again, not the normal way you'd get that sound.

[antti]

Oh, and amplifiers do have "pre-filters". It's called the tone stack.

Wrong. Tone stack is usually (Marshall derived amps) after the preamp where most of the gain and distortion comes from.

There is "pre-filtering", but it's a one-pole lowpass filter per stage caused by the parasitic miller capacitance of the tube. Cutoff is 3kHz or higher, depending on the circuit.

[AdmiralQuality]

Oh, and amplifiers do have "pre-filters". It's called the tone stack.

Wrong. Tone stack is usually (Marshall derived amps) after the preamp where most of the gain and distortion comes from.

There is "pre-filtering", but it's a one-pole lowpass filter per stage caused by the parasitic miller capacitance of the tube. Cutoff is 3kHz or higher, depending on the circuit.

Well on Fender amps it's first. And your guitar has a tone knob too. Anyway, point was, it's not normal to remove large amounts of the highs from the tone before crushing it with pre-amp distortion. (Or after for that matter. Then you'd lose all your crunchies.) But yeah, I don't play that kind of amp (nor do I wanna... blehhhh! )

There are certain tones where it's common for the guitarist to roll off all the highs (usually on the guitar itself), but I think they're more along the lines of sustaining leads, playing higher notes. Not this low crunchy metal stuff. (Sorry if I'm using the wrong words to describe your genre.)

[learjeff]

The hard part about synthesizing the guitar isn't the waveform per se. And while the waveform envelope is relatively complex, that's not the hard part.

The hard part about modeling a guitar is making it sound like it's being played by a human. There is a LOT of technique involved in getting a guitar -- especially an electric guitar fed into an amp with some distortion -- to sound the way they do. I hear lots of folks rave about some player's "tone", thinking it all comes from the gear, when it's mostly right hand technique.

Hot players get hot sounds from virtually any gear, and weak players get weak tone from virtually any gear.

Try modelling that.

When you get it working, then move on to the "Talent" plugin, which I'd really love to have.


What VSTi writers call "physical modeling" is not much like circuit modeling done by electonics CAD programs. They abstract different aspects of the results and do their best to make something that sounds like those results. The methods usually have little to do with the actual physical operation of the thing being modeled.

For example, take a Rhodes electric piano. When the "physical modelers" do it, they realize it has a hammer, a tine, a resonator, etc., and that all those things have specific effects on the resulting tone. Then they come up with waveforms and/or algorithms that seem to sound like the effects of those different things -- but that don't WORK anything like those specific things. That is, when you select "softer hammer", it doesn't cause a physical emulation to change: there's no modeling of the striking of the tine with that softer hammer. Instead, they attenuate the attack or cause other parametric changes to various waveforms that are combined to create the result.

I'm not knocking it, I've played some really nifty VSTi's that use that method. It's definitely more like a physical model than a soundfont, to be sure! But it's not nearly the same as a physical model that's programmed with the hardware details and the laws of physics. As stated earlier, the latter kind won't run in real time on affordable hardware in the foreseeable future, except for modeling the simplest devices.

[satchnut]

Yes.. and thanks, but what's immensely frustrating is that the included song isn't the correct version... if I'm not mistaken, there's a missing sample, and all the Sytruses in that song load the default preset! (I was testing the beta at the time the installer was put together)

Dude, that song was so brilliant I almost cried...And that was the bad version?! Is there a place where I could get the correct version?