Fair question. The upcoming Noah will feature the first physical modeling plug-in of acoustic instruments (plucked string).
Any news if this plugin will be available on the Pulsar/Scope card platforms?
And, better still, any information as to whether physical modeling plug-ins of wind or percussive instruments are being designed/developed/tested?
Any info on this very much appreciated. I am a good PM-fan myself, after all...
I am sure it will come to the Pulsar/SCOPE platform at some point. However I also would tend to think Creamware will keep at least some of the NOAH stuff only for NOAH for quite awhile. This is just my opinion, I would of course love to have PM on my SCOPE.
Indeed... I believe that this modeling era, once implanted and rising, will make samplers obsolete, and I'm looking forward for that... The big issue is the interface which will permit us to operate the virtual instruments. I had a very interesting conversation about that with a guy on the Giga forum, and I'm still wondering how we will interact with physical models. Any ideas?
I would really like to say that Physical Modeling is the "next big thing" in sound synthesis and will win over sampling, subtractive etc. Reality is, Yamaha did a wonderful implementation of several acoustic instruments on the VL1 series but the audience was probably not "ready" to that degree of expressivity.
Now that synthesis exploded thanks to software hosts and DSP boards like Oasys and Pulsar, times are ripe.
I'd like to see if Creamware will be able to pull good acoustic models out of their hats. I have tried Oasys's PM patches and they seemed all but impressive (besides rhodes pianos and percussions and a few basses, the latter two being old Prophecy heritage).
I haven't heard Noah's strings so I can't judge on that. I only know that the huge DSP power of Pulsar might come useful to increase fidelity and the software platform will come useful to increased control.
Just in case you haven't heard it before, there's some examples here http://www.planetz.com/forums/viewtopic ... orum=10&23, and somewhere (couldn't find it rigth now) there's a link to the company that helped with the design with some further early sound-examples.
Agreed. There is no real reason to do away with samplers because they perform more functions than just sample playback of a flute.
Although even with physical modelling in action, I'd still like the flexibility of using samples as well, it helps add texture to what you're doing. Quite often I might have a definitive string sound, but also use a totally different string sample to compliment it in the mix.
The real reason why there are very few Physical Modeling Sythesizers is that Standfor's electronic music laboratory CCRMA is the owner of the patent and they don't want to give anymore for a few coins the wrights for to use it , like they do in the past with the FM patent, (frecuency modulation) with Yamaha.
The ultraexpensive initial price of the VL-1 u$s6000- was the result of this position and not of the cost of the device itself.
As you can see no more companies, includding Yamaha, have released REAL Physical Modeling synths more.
All the designs that are here and there, are not what's patented as Physical Modeling.
This does not mean that thru other ways many designers have arrived to very similar results, mainly in the formant filters design (CW's STS-5000 as an example).
Hi,
True, J.O. Smith's algorithms are patented but the very fact that physical modeling was born out of academic research means the digital waveguide approach (the most computational effective and the only one feasible for realtime polyphonic synthesis on standard class computers) is free and open. I have myself played around with STK (synthesis toolkit), a C++ SDK written by Smith alumnus Perry Cook.
The basics of the synthesis are already there, research is really state of the art. It only takes someone who can invest some serious time on two main issues:
- how do we make realistic sounds, despite of the model inconsistencies (i.e. doesn't matter what happens inside the virtual flute as long as it sounds incredibly close to the real one)
- can we model control on the pretty big number of parameters? can we correlate the bulk of parameters two 2 or 3 only? can we devise a way so that musicians can play a realistic flute or violin only using 2, max 3 MIDI controls.
Another obstacle to the creation of physical models of acoustic instruments is that, put simply, nobody gives a damn of a 100% recreated violin when he can have the Garritans. Plus, the analog/electronic music big trend has pushed researchers to use physical modeling to simulate the discrete circuitry found in analog synthesizers.
Pulsar has a distinct advantage on this. Many of the best implementation of physical modeling were created on DSP systems much less powerful that 6 32-bit SHARCs. It's just about research and good will. I am happy to see, with Noah, the first stepstone is laid. If the sounds are not too bad (at least better than the Tassman's), then I'd say they are on the good track.
You wrote:
"....the most computational effective and the only one feasible for realtime polyphonic synthesis on standard class computers) is free and open. I have myself played around with STK (synthesis toolkit), a C++ SDK written by Smith alumnus Perry Cook."
I'm interested in it, how can I get it?.
Thanks in advance
Pablo
<font size=-1>[ This Message was edited by: PabloFasan on 2002-04-30 21:46 ]</font>
<font size=-1>[ This Message was edited by: PabloFasan on 2002-04-30 21:47 ]</font>
Vanni,
I'm also impressed "STK (synthesis toolkit), a C++ SDK" that what you did mentioned!
Could you tell us more about it? or how do we could get the tool-kit/source-code, that would be nice!
Best regards,
Long
Hi,
I am impressed myself, that my mention of the STK stirred up so much interest In the more general sense, STK is a quintessential modular synthesis system. Sort of Modular 2 in C++ (not Modula 2 ) )
The URL below is where you can download the latest version and try it out. It's a totally open-source project and source can be used and/or modified, provided you credit Perry Cook and Gary Scavone.
Let's not forget CSound for doing a reasonable PM job. The way I use it is in generating various WAV-files, which I play in a sampler. Not efficient, but it's the result that counts
A more efficient way would be to use CSoundVST, which enables you to use CSound orchestras as a VST instrument in your sequencer. I have no idea about it's performance or latencies though.
_________________
ERROR: signature not found. fake it (Y/n)?
<font size=-1>[ This Message was edited by: rvberkel on 2002-05-02 03:54 ]</font>
"...nobody gives a damn of a 100% recreated violin when he can have the Garritans."
Well, as good as it might be, we can say that a violinist can attack a note in 300 different ways. It depends on what one wants to do, but if we talk about fooling the hears of a musician, we must get closer to that point. Also there's the emulation of (de)crescendos which, imho, requires more than a samples patchwork to be close to reality. Sound banks are rapidly evolving towards new heights of quality, but with new generation multi modelers, there's nothing we won't be able to do that we now do with samplers or any other device.
The way I see it, it could be a single unified platform in which we would load different modules with different functions, like the CW platform, only much more complete and with every developer on the scene (every corporation) working on the perfection of this unified scheme. I think it's a shame that all the industry works on such a competitive basis and not a cooperative one, because everybody would gain tremendously from all rowing in the same direction and not trying to have it's own way with the largest possible profits. In fact, with a truly unified convention, every actor could make as much money because everybody's modules could be included in everybody's projects, only what one invents would not have to be reinvented a hundred times because it would be available. These warring economic attitudes really backfire for the evolution of quality tools and rapid advancement of every single area of human activities.
We could give a name to this ultimate platform with ultimate flexibility and quality. How about "Summit"?
Not quite hard-coded, actually. Each synth is implemented as a class but, if you examine the source structure, you will see it does call sub-classes. To leave the programming paradigm, each synth is a module that calls a structure of sub-modules. In case of a flute, filters, non-linear excitator and delay lines are somewhat compulsory. You will be amazed how relatively simple each of these modules can be.
On 2002-05-02 02:05, LHong wrote:
Thanks Vanni for the link.
It seems be very interesting, kinda Hard-coded soft-synth to me!
Best regards,
Long
thanks for the interesting posts. I do agree with you to some extent. Comparing physical modeling and sampling and saying one is better than the other is to me like saying that photorealistic 3D animation (which, like PM, is indeed approaching sense-fouling perfection) is better than photography.
There is and there will always be space for both. With that "nobody gives a damn" I meant producers who just want to get the work done without getting late into the night and drinking too much coffee, need simpler composition tools. This is why PM never smashed in. Next week, I am going to realise one of my most long-cuddled dreams: acquire a used Yamaha VL1-m, for a price that's too nice to be true (especially for a student like me). The very fact that someone is willing to sell such a beauty of instrument means the average composer/producer will never have a perception of the real potential of PM unless the enormous expressivity is unleashed in a control matrix as simple and intuitive as possible. Many says that the VL1-m can give the best of itself only when played with a wind controller and keyboard. I think Yamaha did a mistake in trying to compel users into buying an additional (and expensive) device like the "matching" wind controller.
What about using pithbending (or modulation, hardly both) and mapping dynamics and aftertouch with an intelligent randomisation so as to render each single note timbrally unique and yet consistent and simple to play?
This is to me the real challenge that awaits PM to become the real mainstream of the next, say, 100 years. Technology is mature. Control is less.
But...there is a thriving marketplace for innovative controllers - Noah is definitely one of them. Problem is - how to map controls to synthesis parameters in case of a PM? To give 25 different knobs to a wannabe sound programmer is good for traditional synthesis like subtractive, additive etc. For synthesis methods that breach the cause-effect relationship we are comfy with, namely FM and PM, control mapping is much harder and less and less people are willing to do so. It would be nice if synth manifacturers helped on that by studying and providing in-the-box solution, that people can learn and maybe easily customise at a later stage.
Nice reflexions, Lulai! Your analogy of animation vs real pictures is very relevant, for it illustrates well the present period of history where the level of novelty is so increasing that we do not fully realize its true acceleration. Inevitably, we will get to the point where modeling will become so close to realism that our outer senses will be fooled. When is this going to happen I cannot tell, but I certainly feel this wind of change which throws people into heightened consciousness of themselves and their creative power, and that is good news! Things can happen faster than what we can conceive, in every area of activity, and to continue our discussion, I believe (like many) that the big issue in samplers and virtual instruments is the interface. Let's say we have 150 different attacks, 50 kinds of vibratos, 20 velocities, 50 effects and so on, how will we choose the right note to produce the desired result?
I had a very interesting discussion about that with some people on the giga forums, and a guy had developed interesting ideas on the subject.
This is the link: http://www.northernsounds.com/cgibin/ul ... 3;t=002177
