Physical Modelling Synth VSTs

[cptgone]

Just had the most amazing experience with NUsofting's support and upgrading their wonderful PM synth Modelonia .

Companies simply don't come more customer friendly than this.

Modelonia has been one of my favourite synths for many, many years now (PA would call it a secret weapon , but that seems unsuited for such a labour of love).
A true trailblazing classic

[kraster]

No mention of msoundfactory?

It has a whole host of PM modules and you could even extend that with the ability to build comb filters in Mturbofilter or sculpt your own exciters.

It's easily one of the most capable PM synths out there.

[Dirtgrain]

No mention of msoundfactory?

It has a whole host of PM modules and you could even extend that with the ability to build comb filters in Mturbofilter or sculpt your own exciters.

It's easily one of the most capable PM synths out there.

Well, not readily--it takes more steps than dedicated modeling synths, and in all the videos I've watch of it (the Chandler ones--he rocks), the results haven't seemed better--nor on par. But maybe I missed some videos or examples.

[Ou_Tis]

Just had the most amazing experience with NUsofting's support and upgrading their wonderful PM synth Modelonia .

Companies simply don't come more customer friendly than this.

Modelonia has been one of my favourite synths for many, many years now (PA would call it a secret weapon , but that seems unsuited for such a labour of love).
A true trailblazing classic

If only they would add MPE support... going by the audio examples it's probably my favorite physical modelling synth.

I see they have a new PM+VA synth on sale called Sinmad:

https://nusofting.com/plugins/sinmad-synth

[Jac459]

I think Reason Studios still rules here with Objekt and Friktion. Using objekt as an affect and feeding it any impulse sound is just crazy.

[kraster]

No mention of msoundfactory?

It has a whole host of PM modules and you could even extend that with the ability to build comb filters in Mturbofilter or sculpt your own exciters.

It's easily one of the most capable PM synths out there.

Well, not readily--it takes more steps than dedicated modeling synths, and in all the videos I've watch of it (the Chandler ones--he rocks), the results haven't seemed better--nor on par. But maybe I missed some videos or examples.

There's a whole bunch of presets in Msoundfactory that are physically modeled.

The latest instrument they released, Harper is a physically modeled Harp.

There's the Taiko Drum, Timpani, String theory, Snare maker, Windy and a whole host of others.

If you want to roll your own off the bat, Msoundfactory does have dedicated modules for physical modelling like the Karplus Strong "string" generator module, Impulse train exciter and the resonator and modal filter modules.

And as I said before, if you want to take it even further you can build really elaborate physically modeled instruments using feedback networks with comb filters etc.

I think many of Chandler's videos are more about teaching techniques and principles than achieving an end result.

[k2006]

in free synths,Uhe Triple cheese,Surge with "String" a waveguide-based feedback oscillator ,vcv rack with mutable instruments modules,Lord of Springs,Vital using comb filters.

[DaveClark]

No mention of msoundfactory?

It has a whole host of PM modules and you could even extend that with the ability to build comb filters in Mturbofilter or sculpt your own exciters.

It's easily one of the most capable PM synths out there.

Well, not readily--it takes more steps than dedicated modeling synths, and in all the videos I've watch of it (the Chandler ones--he rocks), the results haven't seemed better--nor on par. But maybe I missed some videos or examples.

There's a whole bunch of presets in Msoundfactory that are physically modeled.

<snip>

Based on all I've seen so far, MSoundFactory is not a physical modelling synth, despite repeated claims that it is, most notably by Chandler ("Chandler guitar" or @Chandlerguitar on YouTube).

First it should be noted that historically a physical model is an actual physical object. Physical modelling in this form is a practice that goes back centuries before the term was used. Many people still to this day regard the term "physical model" as referring to an actual physical object.

Physical Modelling Synthesis from Wikipedia includes "...will typically have several parameters, some of which are constants that describe the physical materials and dimensions of the instrument, while others are time-dependent functions describing the player's interaction with the instrument...." (Emphasis added.)

Also from Wikipedia, but a separate article: "Karplus–Strong string synthesis is a method of physical modelling synthesis that loops a short waveform through a filtered delay line to simulate the sound of a hammered or plucked string or some types of percussion."

The description above contradicts Wikipedia's own definition of physical modelling synthesis and provides a typical example of how the term "physical modelling" has been co-opted and misused. A "short waveform through a filtered delay line" is not a string and does not possess the attributes of a theoretical string such as length, nor is it made of materials (real or theoretical) with parameters such as a speed of sound or a Young's modulus. And of course, Karplus-Strong is light-years away from the historical definition.

Calling a computer program a "physical model" is already treading on thin ice. However, as someone who has worked in computational physics on and off for decades, I don't have a problem with people using the term "physical modelling" to describe computer programs, as long as such programs include parameters that are physical attributes of the physical thing that is being modelled. I do object to co-opting the term to describe computer programs that do not include actual physical parameters.

MSoundFactory is not a physical modelling synth because, from what I have seen, including some of Chandler's videos, there are either no or not enough physical parameters to allow such things as changing material characteristics, string lengths, etc., just as Chandler admitted when I confronted him about it. Chandler seems to have the bad habit of accepting a synth as being a physical modelling synth if it can sound like something physical. That's much too low of a bar for qualification.

I have nothing against MSoundFactory itself, and I especially like the Meldway piano.

---------------------------

One argument that I have seen provided as an excuse for co-opting the term is that the mathematics that results from a physical system can be the same as the mathematics resulting from an alternative system that is not physical. For those technical people without much experience dealing with real stuff, this may seem like a good argument. However, it is not. One can melt ice and produce liquid water. One can condense steam and produce liquid water. Concluding from these two facts that steam is ice is obviously not sound reasoning. Transitivity does not apply because there is not strict equality of liguid water and ice, nor of liquid water and steam. There is also not strict equality of a system, physical or not, and the mathematics resulting from analysis of it; they are two different things, so again transitivity does not apply.

[Ou_Tis]

One argument that I have seen provided as an excuse for co-opting the term is that the mathematics that results from a physical system can be the same as the mathematics resulting from an alternative system that is not physical. For those technical people without much experience dealing with real stuff, this may seem like a good argument. However, it is not. One can melt ice and produce liquid water. One can condense steam and produce liquid water. Concluding from these two facts that steam is ice is obviously not sound reasoning. Transitivity does not apply because there is not strict equality of liguid water and ice, nor of liquid water and steam. There is also not strict equality of a system, physical or not, and the mathematics resulting from analysis of it; they are two different things, so again transitivity does not apply.

Pseudo-physical-modeling seems like a more accurate description.

Does your "not strict equality" argument also applies to the difference between physical modeling simulations and pseudo-physical modeling? Or is there actually strict equality of outputs between a (necessarily simplified) model of a physical system and the "equivalent mathematics" (mathematical outputs?) you're referring to? Also, can the pseudo-physical modeling inputs be either strictly or very accurately mapped to the physical modeling inputs to produce either equivalent or very similar outputs?

[k2006]

according to Urs comb filters like in Zebra are enough to do physical modelling

Comb filters are technically short feedback delays. Just like Waveguides. We just can not call them that because that would infringe patents, brand names or some such thing. Hence you'll also not find any hint at "Physical Modeling" on certain websites. That would probably infringe some brand names as well. I rather think of Zebra's comb filters as little "sculpturing" devices. You catch the drift._Urs

viewtopic.php?t=230895

[DaveClark]

One argument that I have seen provided as an excuse for co-opting the term is that the mathematics that results from a physical system can be the same as the mathematics resulting from an alternative system that is not physical. For those technical people without much experience dealing with real stuff, this may seem like a good argument. However, it is not. One can melt ice and produce liquid water. One can condense steam and produce liquid water. Concluding from these two facts that steam is ice is obviously not sound reasoning. Transitivity does not apply because there is not strict equality of liguid water and ice, nor of liquid water and steam. There is also not strict equality of a system, physical or not, and the mathematics resulting from analysis of it; they are two different things, so again transitivity does not apply.

Pseudo-physical-modeling seems like a more accurate description.

Does your "not strict equality" argument also applies to the difference between physical modeling simulations and pseudo-physical modeling? Or is there actually strict equality of outputs between a (necessarily simplified) model of a physical system and the "equivalent mathematics" (mathematical outputs?) you're referring to? Also, can the pseudo-physical modeling inputs be either strictly or very accurately mapped to the physical modeling inputs to produce either equivalent or very similar outputs?

It appears to me that you are asking about applying an argument to a mere claim. The argument is that a <> b, b <> c, therefore a is not necessarily = c, contrasted with transitivity. How does one apply this to a claim that a = b or a <> b?

If two different physical systems result in the same mathematical analysis, say for simplicity f(t) = sin(omega*t) where omega is fixed, the output of the mathematical model is necessariy the same. This does not imply that the two original physical systems are the same. It may be that one system can have its length changed (therefore changing omega), and the other cannot but may have some other parameter that can be changed (also changing ommega). Physical modelling of the two systems would require two different codes (programs, functions, methods) because f(t) is actually f(t, omega) with two different forms for omega.

So with regards to your second question about inputs: This theoretical mapping that you describe, in order for the program to be classified as a physically modelled synth, would have to at the very least incorporate all of the necessary physical parameters implicitly. For example, length and width might be combined as area in one program and treated separately as length and width in another. If the parameters should actually be treated separately for a more complete physical model, then it could be said that the program that treats area is not as good a physical model as the other - as should be obvious. The problem with being classified as physical modelling synths is that most synths are very far from true physical modelling, but many users have no clue that this is the case because their knowledge of physics and the history of physical modelling isn't very extensive. The same is true of many programmers, especially those who specialize in DSP and digital systems.

[syntonica]

I don't think the actual synthesis behind the curtain really matters, only that if you are recreating the sound of some physical interaction, the parameters are there describing size, shape, material, density, etc.

Regardless, one of the great things about PM is creating instruments that don't actually exist in the real world. Like instruments that belong in a Dr Seuss world...

[DaveClark]

in free synths,Uhe Triple cheese,Surge with "String" a waveguide-based feedback oscillator ,vcv rack with mutable instruments modules,Lord of Springs,Vital using comb filters.

Surge is not a physical modelling synth, either.

"Surge XT is a digital synth and makes no bones about it."

"String oscillator uses two filtered waveguides to emulate plucked or bowed string sounds."

I have no objection to this type of claim, being about emulating sounds and not about physical modelling.

Source: https://surge-synthesizer.github.io/

From what little I know about the other two, they are not physically modelling synths, either.

[pugilistas]

I just want the water....

[kraster]

No mention of msoundfactory?

It has a whole host of PM modules and you could even extend that with the ability to build comb filters in Mturbofilter or sculpt your own exciters.

It's easily one of the most capable PM synths out there.

Well, not readily--it takes more steps than dedicated modeling synths, and in all the videos I've watch of it (the Chandler ones--he rocks), the results haven't seemed better--nor on par. But maybe I missed some videos or examples.

There's a whole bunch of presets in Msoundfactory that are physically modeled.

<snip>

Based on all I've seen so far, MSoundFactory is not a physical modelling synth, despite repeated claims that it is, most notably by Chandler ("Chandler guitar" or @Chandlerguitar on YouTube).

First it should be noted that historically a physical model is an actual physical object. Physical modelling in this form is a practice that goes back centuries before the term was used. Many people still to this day regard the term "physical model" as referring to an actual physical object.

Physical Modelling Synthesis from Wikipedia includes "...will typically have several parameters, some of which are constants that describe the physical materials and dimensions of the instrument, while others are time-dependent functions describing the player's interaction with the instrument...." (Emphasis added.)

Also from Wikipedia, but a separate article: "Karplus–Strong string synthesis is a method of physical modelling synthesis that loops a short waveform through a filtered delay line to simulate the sound of a hammered or plucked string or some types of percussion."

The description above contradicts Wikipedia's own definition of physical modelling synthesis and provides a typical example of how the term "physical modelling" has been co-opted and misused. A "short waveform through a filtered delay line" is not a string and does not possess the attributes of a theoretical string such as length, nor is it made of materials (real or theoretical) with parameters such as a speed of sound or a Young's modulus. And of course, Karplus-Strong is light-years away from the historical definition.

Calling a computer program a "physical model" is already treading on thin ice. However, as someone who has worked in computational physics on and off for decades, I don't have a problem with people using the term "physical modelling" to describe computer programs, as long as such programs include parameters that are physical attributes of the physical thing that is being modelled. I do object to co-opting the term to describe computer programs that do not include actual physical parameters.

MSoundFactory is not a physical modelling synth because, from what I have seen, including some of Chandler's videos, there are either no or not enough physical parameters to allow such things as changing material characteristics, string lengths, etc., just as Chandler admitted when I confronted him about it. Chandler seems to have the bad habit of accepting a synth as being a physical modelling synth if it can sound like something physical. That's much too low of a bar for qualification.

I have nothing against MSoundFactory itself, and I especially like the Meldway piano.

---------------------------

One argument that I have seen provided as an excuse for co-opting the term is that the mathematics that results from a physical system can be the same as the mathematics resulting from an alternative system that is not physical. For those technical people without much experience dealing with real stuff, this may seem like a good argument. However, it is not. One can melt ice and produce liquid water. One can condense steam and produce liquid water. Concluding from these two facts that steam is ice is obviously not sound reasoning. Transitivity does not apply because there is not strict equality of liguid water and ice, nor of liquid water and steam. There is also not strict equality of a system, physical or not, and the mathematics resulting from analysis of it; they are two different things, so again transitivity does not apply.

This is just pedantry.

Msoundfactory is not exclusively a physically modelled synth no more than it's an exclusively subtractive synth, additive synth, sampler or granular synth but it's capable of doing them all

If Msoundfactory can't do physically modeled synthesis then neither can Chromaphone or most of the synths mentioned here.

As I said before there are reams of instruments in Msoundfactory that use those exact same techniques (exciters, resonators and modal filters etc.) to synthesise real instruments.

They are literally there! Actual modelled instruments.

All fully parametric.

You can adjust things like the, overtones, dimension, damping etc. on the drums and you have similar relevant controls for string instruments like Harper or the wind instruments.

Here's the first line of the definition on Wikipedia:

"Physical modelling synthesis refers to sound synthesis methods in which the waveform of the sound to be generated is computed using a mathematical model, a set of equations and algorithms to simulate a physical source of sound, usually a musical instrument."

Emphasis mine.

Physical modelling synthesis int he context of this thread is a METHOD of synthesis. It can be used to simulate real instruments or invent completely new ones.

Ironically Msoundfactory is one of the more capable examples of physical modelling around due to its open ended semi modular nature.