Cleaner poly sound?

Hey y’all

This is a conceptual question I suppose, but perhaps someone can help me specifically with solving this in vcv.

So, my chords sound very “organy.” Not in composition but in texture. I have a feeling that my ear is biased toward organically generated sounds and something like a guitar or piano has so much in the way of shared partials, sympathetic vibrations and maybe phase relationships that my stacked notes in synthesis always sound dissonant, at least in the sense that the partials are grinding against each other.

I’ve been filtering (BRF helps) compressing (poly multi band compression is tough, how do I harsh the fundamental in order to send band cutoffs?) and relying mainly on a sine to create the fundamental and layering busier wave forms over the top.

Should I be trying to lock phases together? I’m resetting the VCO on each note… is this just the nature of synthesis. I think I remember my prophet 5 having this issue but I’m not sure if modern synthesizers deal with it somehow?

Gordon Reid FTW as always. Seriously y’all, if y’all haven’t checked this guys synth series out from Sound on Sound magazine in the early 2000’s…

Synthesizing drums in VCV has become a different thing altogether since I went thru his drum stuff. All articles are available in one PDF somewhere. Brilliant.

Anyway. It seems like I was on track. Maybe. My understanding of this stuff is rudimentary. it seems that a per-voice chorus (surge chorus poly) can really help spread out and fatten up the usual one osc per voice problem I was having. Seems less grindy.

I realize I may be talking to myself.

2 Likes

Yes, the Gordon Reid is a great resource! Creating organical sounds or physical modelling is such a wonderful challenge :slight_smile:

Up to now I prefer to work with (resonating) filters instead of oscillators when I try things out. There are a lot of good filters in VCV, which all behave different and add a different colour to a sound. You can apply the same fm/am/rm, vowel or waveshaping techniques to them or add subtle (again filtered) noise/chaotic modulation as with oscillators. I believe this is also the fundament of modal synthesis (a mode being a single resonator or steep bandpass filter)

Another approach is working with short delays/feedback or what is called Karplus Strong synthesis. You can get really nice (somewhat metallic) string sounds with it.

Combining the two, and you will have a complex of comb filters.

There are some really great modules in VCV rack on all these subjects! Some are excellent building blocks and others are more oriented to a directly usable/musical endresult. From the last I learn a lot by comparing their sounds with my noodles and try to come closer.

Anyway… have fun!

Modulate stuff. Modulate anything. Make the modulation different for each “voice”. Watch Omri’s videos.

2 Likes

you don’t need a per voice chorus. If you are using a “traditional” VCO like the excellent one by VCV, do per voice pulse width modulation. Turn on the sync and do per voice whatever. Or for the totally trivial one from every poly-synth, modulate a filter after the VCO with an envelope generator.

I May have overstated my problem. karplus is cool, but it’s a “busy” sound for me, and you already nailed it the Gordon Reid response, that I am looking for more “organic” (read: simpler) sounds. Karplus’ metallic sound is a bit more than I’m going for. I think I’m imagining something like a rhoades and I know that’s not “organic” but it’s meant to be a facsimile of something “organic”. I need a clean “chorder”.

I haven’t tried the filter approach. Are you saying to use a filter as a self oscillating module?

I’ve been placing band rejects after the VCO with Voct modulating the reject band before an LPF. It helps. Using Bog Audio OSC. Love that thing. I’ll try modulating pulse width. Thank ya.

(The chorus does sound good though, I believe it cuts back on beating?)

Yes I play with the self resonating spot, just below or just above or also just resonating.

I saw once a video of someone emulating a rhodes which sounded pretty good as a fascimile.

I believe you could download the patch in the description. But this was 2 years ago… so maybe the link is not working anymore…

EDIT: This was probably done in VCV1. I think I have the patch somewhere on my old computer. I can look if I can transport the patch to a VCV2 version for you later today.

1 Like

This sounds like just what I need. I’ll reinterpret it from the video. Thanks much!

1 Like

My workaround is to SPLIT poly signals and use entirely different voices to handle each chord note. It’s more work, but it does help avoid the ‘mush’ sound issue. I also like to run the gate sequencing differently per note so things are more arpeggiated and only chordal in the delay or reverb trail.

That said, I only realised the other day that what I maybe should be doing is using an equivalent of the Intellijel Shifty to handle sorting out which note uses which voice.

1 Like

I just opened the VCV1 patch in VCV2 and it still works ok - all the fundamental modules just substituted for the new versions.

1 Like

In an octave-based equal temperament, a chromatic step is the n-th root of 2, where n is the number of steps. Standard tuning is 12-tone equal temperament, so, starting on A 220:

  • A is 2 ** (0 / 12) = 1.0, i.e. 220.0 Hz
  • Bb is 2 ** (1 / 12) = 1.0594630943593, i.e. 233.08188075904 Hz
  • B is 2 ** (2 / 12) = 1.1224620483094, i.e. 246.94165062806 Hz
  • C is 2 ** (3 / 12) = 1.1892071150027, i.e. 261.6255653006 Hz
  • C# is 2 ** (4 / 12) = 1.2599210498949, i.e. 277.18263097687 Hz
  • D is 2 ** (5 / 12) = 1.33483985417, i.e. 293.66476791741 Hz
  • D# is 2 ** (6 / 12) = 1.4142135623731, i.e. 311.12698372208 Hz
  • E is 2 ** (7 / 12) = 1.4983070768767, i.e. 329.62755691287 Hz
  • F is 2 ** (8 / 12) = 1.5874010519682, i.e. 349.228231433 Hz
  • F# is 2 ** (9 / 12) = 1.6817928305074, i.e. 369.99442271163 Hz
  • G is 2 ** (10 / 12) = 1.7817974362807, i.e. 391.99543598175 Hz
  • G# is 2 ** (11 / 12) = 1.8877486253634, i.e. 415.30469757995 Hz
  • A is 2 ** (12 / 12) = 2.0, i.e. 440.0 Hz

A pure perfect fifth is a ratio of 3/2 = 1.5. The equal-tempered fifth is 1.4983, slightly flat. You could keep going: M3 (C# vs A) as a pure interval is 5/4 = 1.25, but equal tempered is slightly sharp: 1.259921.

So your description of partials grinding against each other is… exactly right!

Pure harmonic intervals will be relatively simple integer ratios. The twelfth root of 2 is irrational, so none of the intervals in an equal-tempered scale is pure.

Nysthi has a Poly Scala Quantizer that can use alternate tunings. Some of these will sound fantastic in some keys and terrible in others, so if you match the music to the tuning, you could get a less dissonant result. C major chord → Poly Scala with Werckmeister tuning = much less beating. (Alternately, embrace it in the way impressionist composers did, with blurry fuzzy harmonies.) Or … welcome to the just-intonation rabbit hole…

hjh

2 Likes

I built a module to address that “issue”:

1 Like

Looks interesting, maybe a manual update?

It’s in the library. Is that what you are asking about? Or are you saying that the @Ahornberg user manuals are not very good?

There is apparently no manual and the description next to it in GitHub is incorrect, it goes with the metronome above it. The device is sorta cryptic to me and could use a manual to encourage interest.

Ah, yes. Obviously the manuals are all pretty bad (which is sadly common with VCV modules). I’m sure he would fix the incorrect description if you logged a bug for it.

The only “manual” so far is this video:

3 Likes

Ah, I didn’t realize those were videos! Maybe pretty good for a manual, in that case.

3 Likes

oh, I watched the vid. Hopefully people are smarter than me and realize that those are not still images, but actually links to videos.

1 Like