DSP Robotics Support

We got here an arpeggiator-free emulation of the Korg Mono/Poly, but in full stereo and LP > HP series filters. Also has signal path that can switch from the emulative structure 4-VCOs into VCF (here LP>HP), to 2-pairs of VCOs into two LPFs. Each of the four VCOs plus the Noise can be panned anywhere in the stereo field, so I'm calling it Pano/Poly.
Aside from it's odd paraphony and arpeggiator functions, the real star of the show (IMHO) with the Mono/Poly was the powerful use of it's four VCOs in various combinations of Sync and Cross-mod:
- Sync, where VCOs 2-4 sync to VCO 1 ('Single' mode).
- Sync, where VCO 2 syncs to VCO 1, and VCO 3 syncs to VCO 4 ('Double' mode).
- X-Mod, where VCO 1 cross-mods VCOs 2-4 ('Single' mode).
- X-Mod, where VCO 1 cross-mods VCO 2, and VCO 3 cross-mods VCO 4 ('Double' mode).
- Sync plus X-Mod ... Single mode.
- Sync plus X-Mod ... Double mode.

Wheww ! All this without the complexity of modular patching. Plus these crazy modulations can be turned on and off with the touch of the oddly-named 'Effects' button.
It also had some clever switching for the P-Bend and Mod wheels to make maximum advantage of the Sync modes.



Loaded with over 20 presets, mostly rough stabs at those from the Mono/Poly patch book.
Early Korg synths were noted for their LP > HP series filters (similar to Yamaha's CS-series synths, but in reverse order), so I added a 2-pole HP in series with the 4-pole LPF. The HPF is switched to a second LPF when in '2-pairs into 2-filters mode'.
In the panel space where the Mono/Poly had it's arpeggiator switches, I've put VCF and VCA Velocity controls (the M/P did not have velocity). I've also added a couple of functions that are popular mods/hacks to the hardware (double speed LFO, aftertouch control of mod wheel depth, VCO 4 as option for mod wheel source). Various hardware MIDI controllers can be applied to the VCF and/or VCA (like M/P's rear panel jacks).

There are multiple software emulations of the M/P out there (Full Bucket's is much better-looking than Korg's own, which is cluttered and in-authentic), but none of them has dual filters and per-osc panning (unless Korg's has these as menu-hidden functions, I don't know).

Currently under re-construction.

Micro update -

Since I personally don't care for staring at a lot blue for a long time, I've made the BG color changeable to that of the chocolatey brown of Native Instruments' FM7 (the new white GUI for FM8 is hideous) or the deep, warm taupe of u-he's Bazille.

Also, I forgot to ask in my OP, since this relies so much on sync, why have Martin's oscillators never had sync inputs?

The Mono/Poly is a synth ive always wanted to own. Thanks for making this.

I must have missed this one somehow sorry!

It sounds wonderful and has a great set of presets to demo it. It seems to cover all the bases when it comes to “classical” subtractive synth sounds. Who could want more?

Very well done sir, and thank you for sharing it.

Martin once explained that he didn’t include sync because it would create aliasing. But I do wonder if the sync/reset could be made a little "softer" over a few samples say, to avoid or reduce aliasing. An analogue oscillator would probably never be able to reset in 22uS after all. Martin…?

Cheers

Spogg

I hope the presets are (or can be made to be) somewhat useful - they are mostly based on an old patch book; I have no idea if sound as they're supposed to.

Yes, I've puzzled by the resistance to sync - since one expects it to produce some pretty 'tearing' sounds, and one can close the filter a bit to soften it, a little aliasing doesn't seem to be a big deal.

k brown wrote:since one expects [sync] to produce some pretty 'tearing' sounds, and one can close the filter a bit to soften it, a little aliasing doesn't seem to be a big dea

I'm generally quite tolerant of it, too, but filtering it out when it is offensive can be trickier than you might think. Although it's a bit counter intuitive, an instantaneous jump will create aliases right across the whole frequency spectrum, below as well as above the frequencies that you want, which can be especially noticeable if you have a ringy IIR filter following the oscillator (as most synths do). Hence "closing the filter a bit" might require high-pass filtering as well as low-pass if you want to avoid muddying the low end.

Hmmm - I'm going back to bed, now

k brown wrote:why have Martin's oscillators never had sync inputs?

As Spogg pointed out, naive syncing causes aliasing, To suppress aliasing, you have to take into accont at which fraction of a sample exactly the sync event happens. Analog synths have no problem with that because they work with continuus time.

To make things worse, if the slave's waveform has discontinuities, their alias supression will interfere with the sync alias supression. In the worst case there may be one sync and two discontinuities between two successive samples. This can be accounted for but it is rater costly.

So it essentially boils down to two options:

1. Implement naive syncing as in the stock oscillators. Personally I think it sound so bad, it is not worth the effort.
2, Implement consistent alias suppression for both syncing and the osc's own discontinuities. That is quite a development effort, and the result may be more CPU hungry than the stock oscs.

You see, there is my dilemma.

I guess I just don't get the obsession with aliasing, especially with an aggressive sound like osc sync. The DX7 and the early 4-op FM synths aliased like crazy; didn't stop anyone from using them. When you do osc sync with SM/FS oscs what does everyone find so objectionable - it's just another sound color that may or may not work in given musical context - - or am I just deef? It's a mild bit of sync that makes the acoustic piano preset in my Prophet T8 emulation sound so realistic; can't get that sound any other way. Sure when over used (too great a pitch spread between sync'd oscs, it's pretty ugly, but at reasonable spreads it has it's uses.

I vote for both naive syncing and a more CPU-intensive consistent alias-suppression- but then, I'll not be the one having to code them, will I?