Ideas on acoustic models and synth

[Magpel]

For a current project, and general interest, I'm working on coming up with distinctive lead voices that layer acoustic physical models from my Z1 with synth lead sounds. Looking to spark some discussion on good ways to do this.

 

My first goal had been to chose a model--say the Z1's clarinet which sounds decent but wouldn't fool anyone--then try as closely as possible to duplicate the timbre on a VA. Of course you can't get very close, but that's the point. I strip all FX off both sounds. Then I try to get the envlelopes and filters more or less responding the same--again, a little difference is inevitable and desirable, but, say a dramatically different release setting is unnacpetable...

 

...Because, I want the layers to act, play, and sound like one voice. So far, I've come up with some rich timbres but nothing that is satisfactorily intermingled into one voice.

 

I've had a few ideas, like using some complementary EQ niches in each voice to join them better. Haven't gotten very far with this yet because the softsynth VA I'm using--the Pentagon I--has very limited onboard EQ.

 

Sometimes, it seems to be primarily an issue of finding the right volume balance and making sure than any variations caused by velocity or aftertouch are musical and compatible.

 

Has any tried this kind of thing? Any ideas on good combos? My current project is to essentially deface the clarinet tone while keeping its modeled characteristics on the Z1. I want to try the same approach with the Z1 Violin, which is not all a convincing model sonically, but does have a variety of cool expressive devices programmed, and some neat modeled timbral characteristics. Again what I want is a new sound that plays somewhat like the violin...

 

Just groping.

[Tusker]

Nice ideas John. Let me offer couple of thoughts and see if they are any use.

 

The best glue for joining two different sounds together is to send them through the same effects bath.

 

When layering it helps me to think of horizontal and vertical layers. On a horizontal layering strategy, you would be very aware of which aspects of the frequency spectrum are being primarily defined by each voice. With a vertical layer, one voice could add character to the attack and another voice could dominate the sustain/release portion of the note.

 

With horizontal layering, I'd be tempted to tune the acoustic model an octave or a fifth (or an octave and a fifth) above a fundamental derived from the analog model. Reason? I imagine the burps and wheezes of the acoustic model may sound awkward below a constant sounding analog sound. Having an analog sound providing a rich smooth continuity may reduce the "plasticness" factor of the physical model. I am guessing. The reverse might work too.

 

In the vertical layer, I'd be tempted to get the attack from the physical model, particularly if there are some controllable pitch articulations on the attack. The heightened clarity of the physical model seems to suggest that too.

 

Best Wishes,

 

Jerry

[RABid]

Interesting. I guess as an owner of Pentagon I you got the notification about z3ta. I got the full version and played with it last night. There are a lot of possibilities. Last night I got into a tangent of programming control of overtones and ghost voices that would blend and fade on command. My goal is to first create a nice expressive lead that is evident when someone first listens to a song, but also have some controlled voicing that is only apparent when you really listen closely. Sort of that Steely Dan effect where you think the song is simple but for some reason you cannot play it the way they do. I want the same thing out of my lead sound. Simple, yet something else there that other musicians will pick up on.

 

Robert

[Magpel]

Thanks, Jerry, for some great ideas. Robert, I did get the notification regarding the new RGC instrument. I'm in a strict "No Buy" zone at the moment, so I chose not to go read more or check the discussion at KVR. Avoid temptation, you know.

[aeon]

Magpel said:

Haven't gotten very far with this yet because the softsynth VA I'm using--the Pentagon I--has very limited onboard EQ.

Pentagon I may have a signal flow like a VA, and indeed have plenty of virtual knobs, but make no mistake...it is not a VA. Pentagon I uses PCM samples with transposition and interpolation to generate the synth voice, so it is a soft-ROMpler.

 

As far as blending two synth voices from dissimilar instruments/engines in order to create a singular voice, I have gotten great results from passing a mix of both signals through a spectral shaper of some sort, e.g. Antares Mic Modeler, or a complex resonance curve built in a FIR filter environment.

[Tusker]

Originally posted by aeon:

As far as blending two synth voices from dissimilar instruments/engines in order to create a singular voice, I have gotten great results from passing a mix of both signals through a spectral shaper of some sort, e.g. Antares Mic Modeler, or a complex resonance curve built in a FIR filter environment.

Aeon:

 

That sounds cool. Where can I get more info about FIR?

 

With my lowly onboard synth efx, I find that the exciter algorithms are a better glue than the typical reverb and chorus type stuff.

 

Cheers,

 

Jerry

[aeon]

Tusker asked:

Where can I get more info about FIR?

On the Mac, I use an older VST plug-in called FireBall, from Intelligent Devices. I dont think they offer it any more, as the info is gone from their site.

 

It offers you a 4096-point Finite Impulse Response filter/EQ, so you can build complex resonance curves, but have zero smear, phase problems, or ringing.

 

http://www.tweakhead.com/fireball.jpg

[Synthguy]

We need more threads like this. And participation.

 

Sort of like Magpel, I'm exploring the world of synthesis a little more now in trying to do things the old school way, when people only had analog beasts and tried to synthesize sounds and textures that sometimes imitated acoustic ones. Later with better synthesis technology, they worked on giving some sounds more acoustic characteristics, like they were playing physical models.

 

To me, Tomita is still the master of synthesis, having basically covered most of the sonic landscape, and with his Modular Moog system managed to capture timbers that people with the most advanced studios are still only matching. I know that todays technology allows the creation of all sorts of new classes of sounds, but it's rare that I hear something that Tomita hasn't created first, almost 30 years ago.

 

I've become fascinated with the potential of using these newfangled synths, not to create the perfect illusion of a Steinway or symphony, tho it's easier every year with the quality of samples in the new Yamahas Korgs, Rolands and Kurzweils, not to mention Gigasampler. Instead, I want to create as many sounds and textures with synth waves and the synth's engine and effects, using no acoustic samples, and see if I can use my synths like they used to be in those electronic albums of yore. I'm considering going so far as to take some synths and combine them in teams, so to make patches I use one layer from a Korg, and one layer from a Kurzweil, treating them like a single instrument, using their sonic character to craft a hybrid.

 

Has anyone done or heard of much along those lines?