Separating oscilators and filters in their own vst

VST, AU, AAX, CLAP, etc. Plugin Virtual Instruments Discussion
RELATED
PRODUCTS

Post

In my quest to get as much one-page interfaces as possible, I'm considering using the oscillators on one vst, and then filters on another. In my case, zebralette and tone2 filterbank. Zebralette shows visually what's going on with osc waves, and one envelope (the lfo is not visualized). Filterbank shows fantastic visualizations on what's going on with the filters.

They are both mostly one-page apps.

My questions...
  • Zebralette is only one OSC. A very powerful one. Why do we need more than one? detuning and unison are of course possible with Zebralette. Or am I missing something?
  • I won't have presets. If I like a sound, I'll have to somehow hack together the patch names on both Zebralette and the filterbank. Maybe using minihost or similar would work, but then the DAW cannot map controllers to CC properly. Any workaround?
In general, is this a good idea at all? Not going for zebra straight as it has a multipage interface, and it's too complex for my goals (learning visually synth programming).

Post

You could load both in a modular environment, for example MUX. In MUX you can load VST instruments and effects, as well as the MUX modules and connect them together. Then you can save the settings as ONE preset, so you don't have to save the Zebralette preset and the Filterbank preset individually.

Post

MUX I tried, but it breaks 'control link' (the way to map CC's to knobs in S1). So not usable. Same for minihost.

Post

Maybe with a different DAW?

Post

urlwolf wrote:In general, is this a good idea at all?
Nope.

Firstly, this would mean that you have no way to do polyphony properly. There's no mechanism for splitting different voices out for processing independently, so, for example, if your 'oscillator' VST produces a chord, you'll get a single filter on that chord rather than one filter per note.
Secondly, the way hosts work is that they don't send a continuous stream of samples, one at a time, through each plugin in a chain. Instead each plugin gets sent a buffer, say of 32 or 64 samples, which it then processes in full before the buffer gets passed on to the next plugin to deal with. That's going to have an affect on all sorts of things; you're potentially talking about a couple of milliseconds between a sound starting and being filtered, for example, instead of a couple of samples.
Thirdly, you lose all sorts of possible interactions eg filters being frequency modulated by oscillators. Or tremelo slaving off the same modulation source as vibrato.
my other modular synth is a bugbrand

Post

Thanks whyterabbyt, makes sense. Glad I asked.
So what's the point of things like filterbank? Used as an effect?

Post

well it's not a bad idea… it just has limitations, which is exactly what you're going for isn't it?

Post

Bidule?

Post

The 'why do we need more than one osc' is also an interesting question.
I guess a single one, but capable of producing a shitload of waveforms, is easier to learn than the complex interactions between 2, or 3, or 4 oscs... even if the waveforms are simpler?

Post

urlwolf wrote:Thanks whyterabbyt, makes sense. Glad I asked.
So what's the point of things like filterbank? Used as an effect?
To filter other audio sources, including synths that have their own filters. I almost always have secondary filters on buses. They're not to shape the primary synth sounds but to apply as an effect.

It's what you're hearing when entire portions of the track are being filtered.

Post

urlwolf wrote:Thanks whyterabbyt, makes sense. Glad I asked.
So what's the point of things like filterbank? Used as an effect?
Well, that'd let you filter the timbre of the track as a whole, which is a relatively common thing. Its not the case that thats only going to be done to a synth, btw, but in some cases you might want to...

Think of it this way; within a synth filters are used per-note to shape the evolution of the sound of that note. That's analogous to, say, a guitar, where the way you play each string (and its physical/magnetic interaction with the pick, guitar, pickups) also determines the evolution of the sound of each note.
Then compare that to running that guitar into an amplifier. The amp modifies the entire sound of the guitar.
Both 'stages' (the per-note interactions, and the global processing of the whole guitar sound) have their place in the final result.
That's sorta why you might still want to process a polyphonic synth with per-note filters with a single external filter operating on the whole output. You want to control the timbral evolution of the notes individually, but also apply an effect to them all as a single instrument.
my other modular synth is a bugbrand

Post

urlwolf wrote:The 'why do we need more than one osc' is also an interesting question.
I guess a single one, but capable of producing a shitload of waveforms, is easier to learn than the complex interactions between 2, or 3, or 4 oscs... even if the waveforms are simpler?
You need multiple oscillators if you want to use FM or ring modulation. Or to play intervals (fouths, fifths, & octaves, frequently) from one key.
I wish I could sing as well as the voices inside my head...

http://www.cdbaby.com/darkvictory

Post

urlwolf wrote:The 'why do we need more than one osc' is also an interesting question.
I guess a single one, but capable of producing a shitload of waveforms, is easier to learn than the complex interactions between 2, or 3, or 4 oscs... even if the waveforms are simpler?
It might be easier to learn, but that's not sufficient to only utilise single oscillators, IMO. As DocAtlas says, there are synthesis methods that rely on interactions between different oscillators, eg FM, sync, ring modulation, plus you have the ability to stack multiple sources at different frequency intervals. The results of these are not necessarily reproducible in a single oscillator; they can introduce enharmonics and dynamic timbral changes that can, at best, only be approximated with a single oscillator, and at worst, cant be done at all.

If you had two oscillators, and the second was totally limited so that it could only ever be set such that its base frequency was an integer multiple of the first oscillator, and its output amplitude also only had a specific number of possible options, then that could be replaced by a single oscillator with more waveshapes. Except to do just that, you'd need to provide many more waveshapes.
In fact if you had X waveforms in oscillator 1, you'd have to provide

X * (number of waveforms of oscillator 2) * (number of frequency options for oscillator 2) * (number of amplitude options for oscillator 2)

waveforms.

However if oscillator 2 has full control over its frequency and/or amplitude, then you lose any possibility of replicating that exactly from one oscillator.

Now, its entirely likely that not every oscillator combination sounds different enough from others to necessitate a unique option in your single oscillator. But even so, the number of possibilities is vast. Vaster than makes a single oscillator the 'easier' option.

And that's all before you think about oscillator-level interactions like FM etc.
my other modular synth is a bugbrand

Post

Good to hear.
This makes sense. Better learn 'classical substractive' rather than the suppa-duppa oscillator from zebra.

Then my learning synth will be Photone. I really like the way it sounds, it can do decent FM, and the envs have some animations. The filters it has, no idea what they are doing, and nothing in the docs. I like how the mod matrix is arranged: dropdowns on the GUI element it modifies. not many things do that.

Seeing the oscillator wave is kind of a luxury. The envs and filters, kind of necessary to learn.

Post

Dasheesh wrote:Bidule?
Yes.

You can set something like this up in Bidule, but it's not going to to be terribly efficient. You'd have to set up a mono- or duo-phonic chain, replicate for the number of voices you want, create the control connections, and use the note splitter to dispatch notes to each instance (which doesn't currently have any options for note priority or dispatching order ). It would be complex to set up and manage, and it would eat lots of CPU... But it would be possible.

The idea is interesting and I remember thinking it could be realized with a virtual synth rack that would be a common open standard,like the 500-series racks used for processing hardware. Then different people could contribute their own versions of various processing modules based on a common framework of interconnects.

But in the end, ours is a very small market and an idea like this isn't really economically sustaInable. In its stead we have the current crop of advanced modular synths and processors (Diva, Reaktor, Molecular, etc...), all bound to the same dev universe by necessity. For the more adventurous, there's Bidule, Mux, Usine, etc... But I agree with the OP that it's an interesting idea.

Post Reply

Return to “Instruments”