JUPITER-80

Physical Modeling (This is cool)

Physical Modeling (This is cool)

2 Feb 2013, 06:43
Well I was using my Jupiter 50 (same engine as Jupiter 80) and, since, people talk about whether it's physical modeling or not...I did a test.

I think that the supernatural acoustic section might be sample based. But I think the Jupiter 50 synthesizer section (even the PCM waveforms) is physically modeled.

Basically, a rompler uses samples and then shifts the pitch. If something is generating the sound, it changes the pitch without making the sound go faster, or slower, unlike a rompler. I realized that the Jupiter 50 pitch bender is stretching the pitch, like a rompler would stretch a PCM waveform. So, I went to the sounds and transposed the pitch 24 up and I shift the octave 3 octaves higher. Now I could easily notice samples (I felt like I was listening to Skrillex for an hour). Anyway, I figured that, since, a rompler takes a samples and shifts it...using the pitch bender, shifting the note and then pressing a key (of the note i shifted to with the bender) should be the same sound. I used the pitch bender from up 1 - up 12. Anyway, then i tested it, the sound was different with the pitch bender as opposed to the note. For example, shift up 5 and then press 5 notes higher and the sound is different. This would mean that the synth is generating the sound using va.

Does this makes sense?

Re: Physical Modeling (This is cool)

2 Feb 2013, 12:38
Yes, the synth engine in the Jupiters and Integra are definitely virtual analog. On a rompler, you can tell that the synth waves are samples when you do things like you posted. On a VA, the oscillators and its waves are generated in the engine, and calculated to create all the dynamics of the sound. On a good one, it's hard to tell the difference between an analog synth and a VA, and the Jupiters are superb.

Re: Physical Modeling (This is cool)

2 Feb 2013, 14:49
Physical Modeling IS NOT Virtual Analog ! It's more complex way than signal processing for sosutractive synthesis. Physical Modeling Synthesis can afford every type of sound generator, Vitual Analog can't ...
Vitual Analog can be translated by "Non-Analog Soustractive Synthesizer" and the worse things are in these categorie ... Only few of them have a taste of analog flavour ...
But, OK, you're right ! Jupiter is just another VA asshole , physical modeling is just a marketing joke ...

Re: Physical Modeling (This is cool)

3 Feb 2013, 04:05
Physically modeling is simply the idea of trying to use software to model physical properties of sound making devices by emulating their actual physical properties and of various components, so that the resulting sound results from these emulated physical components, as opposed to trying to "synthesize" the sound you want.

The idea is to break it down to the physical parts and components that make up the sound creation and virtually emulate these components, down to the way they vibrate or whatever.

Synthethis, is when you use various forms of synthesis, such as subtractive, or addictive, or FM or whatever....and attempt to create sounds that way. When you make a violin sound with a synthesizer of any kind, you are not modeling the physical properties of a vibrating string on a wooden violin body, etc.. Rather you are using oscilators and filters to attempt to sculpt a similar tone using an entirely different process.

So a physically modeled violin would attempt to create the violin sound by attempting to emulate the actual physical components of a violin, the way the string vibrates, how thick is the string, how big is the resonating violin body, what kind of wood is it made out of, etc.

Now then, a real analog synthesizer also has physical components. It has an osculator. It has filters, etc. These electronic components interact and create sound. These physical electronic components can be physically modeled absolutely! is that what is going on in the Jupiter, that I'm not sure, but if they call it VA that is normally thought of as something along the lines of physical modeling because they are breaking the software emulation down to very low level, trying to emulate the physical properties of a real analog machine.

The Clone wheel organs also could be considered physically modeled to a certain degree.

Re: Physical Modeling (This is cool)

3 Feb 2013, 18:22
http://en.wikipedia.org/wiki/Physical_modelling_synthesis

Quotes from JUPITER :: Roland Synthesizer Global Official :

Authentic synth sounds — from classic vintage analog to modern digital

To faithfully recreate the unique sound of analog synths, we've meticulously analyzed the behavior of oscillators, filters, and envelopes that are unique to analog synths. In the new JUPITER series, a single synth tone comprises three OSC, FILTER, AMP, and LFO chains, which give you absolutely powerful synthesis control over a single tone. You also have access to ring modulation, WAVE shapes for creating complex tones, and the popular UNISON mode, which lets you overlay up to eight waves per OSC to create thick and fat analog synth tones.


Image
Incredibly Smooth Analog Feel
Emulates the envelope behavior of an analog synthesizer
In an analog synth, the electric charge/discharge rate of the envelope rises and falls in a slightly curved line. The JUPITER-80 and -50 faithfully reproduce this attribute for all of its envelopes, including the oscillator, filter, and amplifier. By using this envelope to control the four LPF types, the resulting sound has an incredibly smooth analog feel.



Image
Emulates the filter behavior of an analog synthesizer
The LPF plays an important role in shaping the tone. It can be used to cut elements from the sound, and can also be used to emphasize elements around the cutoff frequency with resonance. When you control the resonance in an analog synth, the cutoff frequency fluctuates, and signal levels below the cutoff frequency are affected as well. The entire filter changes organically as if it's alive. These changes occur due to the behavior of the analog circuits and the components used, such as resistors and condensers, which make up the unique sound of an analog synth.


Image
Four types of LPFs with unique behavior
[JP-80, Jupiter-8, MiniMoog, Prophet-5]

The JUPITER-80 and -50 feature four types of LPFs (JP-80, Jupiter-8, MiniMoog, Prophet-5). Each filter behaves differently, even if their cutoff and resonance parameters are the same. The LPF is personalized by its parameter settings and real-time control cutoff frequency and resonance.

Analog Feel 0–127
Use this to apply “1/f fluctuation,” a type of randomness or instability that is present in many natural
systems (such as a babbling brook or whispering breeze) and is perceived as pleasant by many
people.
By applying “1/f fluctuation” you can create the natural-sounding instability that is characteristic of
an analog synthesizer.

Re: Physical Modeling (This is cool)

11 Feb 2013, 09:57
Nice colours on those examples Dany. There's my colour therapy for the day...now on to some aromatherapy!

Who bloody cares whether the BMT and SuperNATURAL technology is PM, VA, LA, subtractive, additive, (loved the one that called it "addictive!" ;-D) manipulative, creative, or destructive? There was a video that exhibited how the BMT was not using multisamples, and the pitch bend displayed a smooth rather than a stepped response (in this case on a piano sample).

OK, so the analog purists can probably find fault with every VA machine out there, but to my ears, and within a mix it is close enough, and the SN sounds of the brass like trumpets and trombones and strings like violins and cello's just thrill.

As for the Jupiter-80 being just another VA a-hole...yeah well, everybody has one!

as in: quot homines...tot sententiae!

Re: Physical Modeling (This is cool)

11 Feb 2013, 11:52
Michael818 wrote:
I think that the supernatural acoustic section might be sample based. But I think the Jupiter 50 synthesizer section (even the PCM waveforms) is physically modeled.

My understanding is that the synth section lets you choose from 7 modeled analog waveforms, and 350+ PCM waveforms. By common definition, PCM = sampled. So if they are modeling the PCM waveforms, they're using confusing terminology. Not that that would be a first for Roland. ;-)

Re: Physical Modeling (This is cool)

11 Feb 2013, 12:53
I seriously doubt they are "modeling" the PCM waveforms. They are very complex waves. Modeling a sine wave, square wave, sawtooth, etc... to mimic an osc, is one thing and not hard to do. Trying to model a sampled wave is an entirely different matter. Model how? No those are not modeled.

However, everything after the PCM wave, when used in the synth, would be going through other VA elements, such as the filters.

As to why the pitch bend sounds different then the note pitched up the keyboard, I would say that has more to do with the fact that when you pitch bend a sample, is the same sample being bent up. When you go up the keyboard, there are other samples for different key ranges, etc..

Re: Physical Modeling (This is cool)

11 Feb 2013, 13:41
Its not that hard to comprehend but some folks just can't miss taking a negative shot at Roland. PCM waves are samples, not modeled but they are processed by the SN synth engine which has modeled filters.

Re: Physical Modeling (This is cool)

14 Feb 2013, 09:27
Maybe. I don't know as a fact. But i just thought that Roland might be using resynthesis (which is, actually, easy to model). I like that diagram of the va that was posted. Very detailed

Re: Physical Modeling (This is cool)

14 Feb 2013, 15:17
Good info from an early JP-80 thread on BMT right here on RCF:

http://forums.rolandclan.com/viewtopic.php?f=44&t=37058&start=15

I think that the bottom line is that SuperNATURAL and BMT is NOT Physical Modeling per se, but an evolutionary progression of the more expressive technology that started with the ARX and AP-S on the Fantom and V-Synth's.

Re: Physical Modeling (This is cool)

14 Feb 2013, 22:24
SoundworldA.D. wrote:
Good info from an early JP-80 thread on BMT right here on RCF:

http://forums.rolandclan.com/viewtopic.php?f=44&t=37058&start=15

I think that the bottom line is that SuperNATURAL and BMT is NOT Physical Modeling per se, but an evolutionary progression of the more expressive technology that started with the ARX and AP-S on the Fantom and V-Synth's.
As I understand, the Roland promo term "SuperNATURAL" means a combination of different technologies, like APS (Articulative Phrase Synthesis), BMT (Behavior Modeling Technology) and Physical Modeling.

The SN Synth engine seems to be the only one, which clearly involves Physical Modeling, like the above figures, explanations and the modeled Jupiter-8, MiniMoog and Prophet-5 filters indicate.

The JP-80'a SN Acoustic section offers a few APS sounds (like e.g. 0097: APS Trombone), which use a simple synth waveform on which APS of the corresponding acoustic instrument is applied.

Otherwise, all the other SN Acoustic sound engines involve samples, BMT (which seems to be an extension of the original APS ) and partially some kind of Physical Modeling, as stated by Roland:

http://cms.rolandus.com/assets/media/pdf/jupiter-80_tech_brief.pdf :

"...Each SuperNATURAL instrument—piano, violin, trumpet, etc.—has its own specialized engine. That’s because, just as the sounds of a piano, flute and tympani are very different, so are the attendant modeling needs...

...Each SuperNATURAL engine is indeed based on top-quality studio samples of the
original instrument, however that’s where the similarities to PCM-based gear stop.
For starters, SuperNATURAL does not employ sample looping—resulting in smooth,
organic decays. Moreover, Roland engineers developed technology that creates
models of the formative characteristics and responsive aspects of each instrument.
The SuperNATURAL engine transparently handles how those parameters react to
performance dynamics and nuances from a controller such as a keyboard—just as
they would in the original instrument...

...In perhaps the most sophisticated example, the SuperNATURAL
piano engine in the JUPITER-80 allows players to easily adjust the desired amount of
string resonance, key-off noise, hammer noise, stereo width, tonal character and more.
These continuously variable modeled characteristics in SuperNATURAL engines are
a sharp contrast to what are largely snapshots of timbres in electronic instruments
relying solely on samples. The upshot is a distinct leap in realism with respect to both
sound and responsiveness..."


http://www.roland.com/piano/SuperNATURAL_Piano.html :

"...This natural expressivity was our goal in the development of the "SuperNATURAL Piano sound engine." Through Roland's unique V-Piano technology, we have been able to achieve natural and seamless variations in tone like that of acoustic pianos...

...The "SuperNATURAL Piano sound engine" changes all this. With its use of V-Piano technology, it produces a natural decay in which its tone gradually changes as it decays, as it would on an acoustic piano..."


As the V-Piano is based on Physical Modeling, the above statements from Roland seem to indicate an involvement of some Physical Modeling technology in the JP-80's SN Piano engine. Roland of course doesn't reveal the details...

-

Re: Physical Modeling (This is cool)

15 Feb 2013, 02:49
lol, i guess not too much detailed information about the engines are available. That tech brief did explain a lot, though. Sample resynthesis is pretty common, though. Izotope Iris and Camel Audio Alchemy are a couple ones. Nonetheless-it's still pretty interesting doing research about the Supernatural and BMT technologies :)

← Back to JUPITER-80