The boundaries of Synthesis?

[sprech_stimme]

 

Originally Posted by TechEverlasting The emulation of acoustic instruments and creation of new "virtual" instruments is really just beginning. The dichotomy between "sampling" and "physical modeling" is nonsense, these should not be separate approaches. - Once we have powerful enough processors and adept enough programmers to process sampled waveforms in ways that model the way real acoustic objects behave we'll be able to start having some real fun.

 

 

 

Isn't this the approach that Spectrasonics Omnisphere is trying to do?

 

 

Uh - no. Omnisphere sounds great, but is essentially a large sample library bundled with a sample + subtractive synth engine.

[setAI]

physical modeling of organic and acoustic sources is still just emerging- not for reproducing instruments but for simulations of animal vocal tracts and things like complex hydraulics-

oscine tract and sysnsect are good examples of a start: http://www.breathcube.com/vst/nature/birdsandinsects.html

[Tusks]

its been available for quite some time. but its marketed as reverb, and for some odd reason most people cant see past the name and use it only as such. try googling "convolution reverb hardware"... its really just the same problem as with "physical modeling" synthesis, which is all just IIR filter systems. manufacturers think everyone wants a more realistic piano or hall reverb, but there is really a lot of fun to be had with these things..

 

 

I haven't heard of a convolution reverb that lends itself to real time (midi or cv) control. They allow sound to processed, but not performed. Useful and fun, but not musical instruments.

 

I am hoping you are aware of specific examples and that I am incorrect.

[realtrance]

While we're all dreaming of more more more, I'd say the single biggest limitation to the current range of synthesis techniques is that.... the massively vast almost completely 99.9% majority of synthesists don't even begin to use 'em.

 

Companies over the years have banged their head against that brick wall with even modest attempts: Yamaha VL (mass-market implementation of physical modeling), Yamaha FS1R (Formant synthesis with mega-FM implementation), vocoding (hell there's a lot more you can do than Robot Voice and Pitch-Harmonize with this alone! and yet those two are again the current fad), Microwave XT & Q and Blofeld (wavetable synthesis), Virus TI's and Absynth's and Massive's implementations of granular, and the current king-daddy of them all, V-Synth/GT -- to name a few examples -- and yet the reactions have been, from a sales and even "innovative patch collections from users" point of view, almost nil. Sure, there's some initial flurry of interest, but the fact is, the vast majority of players either don't understand or don't want to understand the details deeply enough to really innovate with them, and thus we end up with, "well the FS1R makes some really nifty basses!" and "the latest v2.0 patch set for V-Synth sounds a LOT more like the good old JP-8000, Thank God Almighty!" etc. -- and on and on.

 

How many amazing collections of mind-blowing granular patches for the latest Virus OS are there out there on the net? Two, maybe? If you're lucky? And you can't tell me that's because the truly innovative patchers are just keeping all their incredible work to themselves. Not believin' it.

 

Sigh.

 

Then the other side of it is, there's a level of understanding of both the fundamentals of music _and_ the fundamental electrouacoustic principles of sound generation that subsets the tiny minority who're interested in innovating with new synthesis techniques yet again. V-Synth and VirusTI probably do the best of trying to make some of the basics of "different forms of synthesis" very performatively accessible, but even there, you really have to know what you're doing, it's not just a question of "twisting knobs until it sounds cool" (trying to take more complex forms of synthesis and reducing them to that kind of control is indeed one of the innovations of the last few years).

 

So, there's maybe six people on the planet who have the musical as well as electroacoustical knowledge _plus_ something other than a completely deaf ear (academia is rife with legions of people who've mastered all the technique in the world but just end up with a slightly more interesting "bloop!shhhh!" at the end of it all) who can do something with all this.

 

And maybe twelve people who can do interesting things with basic, two-oscillator analogue synthesis, other than imitate the Six P-Funk Bass Sounds and Four Depeche Mode Synth Sounds From the '80's as the epitome of their technique.

 

Double-sigh.

 

Back to the darkness from whence I've come!

[jskpongoui]

There are no boundaries, with hard work and patience anything is attainable.

 

[Zentrails]

"While it may be argued that sampling is not synthesis, "

Not if you have a Yamaha SY77 or 99.

I'd like to see chaos theory used with physical modelling as a starting point, that would be interesting.

[Zentrails]

"I haven't heard of a convolution reverb that lends itself to real time (midi or cv) control. They allow sound to processed, but not performed. Useful and fun, but not musical instruments.

I am hoping you are aware of specific examples and that I am incorrect."

FDSP (Formulated Digital Sound Processing) Synthesis as in a Yamaha EX-5

[Meatball Fulton]

realtrance's comment is dead on. With all the synth power we have at our disposal today nothing really new is being done with any of it. (I am as guilty as chanrged here...)

Also the earlier comment about expression, I've said this here many times but I have more control over the final sound with my fingers on the strings of my bass than I do with banks of knobs on any synthesizer.

[plaid_emu]

Goofball Jones wrote:

In my opinion Physical Modeling is still in its infancy. There is a big difference between making a calculation that produces a signal that sounds like a physical instrument (e.g. Karplus-Strong) and generating a signal using a mathematical model based on first principals. At some level it may be possible to make an isomorphism between a calculation such as Karplus-Strong and a numerical method used to estimate solutions to a mathematical model involving a very simple Wave equation, much like one can make analogies between various network/node equations in electronic circuits and structures with numerical methods for solving the Laplace equation. However there is a big difference between that and developing a full set of equations for the generation of sound by a specific instrument.  As an aside it is possible that a specific model for a given instrument is not required. For example two drums with different drum head shapes may produce the same sound (spectrum). There is a very famous mathematical paper entitled"Can one hear the shape of a drum?" by Mark Kac (American Mathematical Monthly 73 (4, part 2): 1

[Conbrio]

---

realtrance wrote:

While we're all dreaming of more more more, I'd say the single biggest limitation to the current range of synthesis techniques is that.... the massively vast almost completely 99.9% majority of synthesists don't even begin to use 'em.

Ah yes, quite true. 

[Rudolf von Hagenwil]

The next big thing will be that all you guys and gals will all be replace by LVB SYNTHESIS.

 

And my next big thing in synthesis will be that I throw all that stuff in the garbage and learn to sing like a black bird and a waterfall, and live in solitude with 40 nuns.

[wwwjd]

Why haven't they invented a new GUITAR also?   I mean, this whole 6 strings on a fret board is getting a little long in the tooth.  They need to keep advancing that and making it new and better every year.

Or a better acoustic piano maybe?  New forms of violin?

...or, wait.  Maybe they don't have to?  

Discontentment is more profiable than contentment.  So we are bred to beleive bigger, faster, newer is better when contentment is all we really want.  Yeah.  Ponder that for your next A.D.D. minute.