Anyone played a synthesizer polyphonically?

[Tusker]

I'd like to find out if any of you have played a polyphonic synthesizer (as defined below):

 

A polyphonic synthesizer is one where multiple notes interact which each other, and influence each other. The synthesist is provided programming tools to define how these interactions will affect timbre, pitch, tone duration, etc. i.e. a synthesizer which has many voices that act independantly from each other (xp80, s80, Triton, erc) would be multi-monophonic, by this definition, not polyphonic.

 

To be fair the effect sections of some synths allow voices to interact, but rarely do voices interact in the synth architecture itself.

 

Crude examples of polyphonic behavior:

 

- When one plays middle c monophonically, on a grand piano, it sounds different from a middle c played as a part of a chord, due to resonances between strings.

 

- When one plays more than one note on an overdriven amplfier (typically on an electric guitar) the pitches interact in unpredictable ways sometimes.

 

I'd like to find out if anyone has ideas for polyphonic synthesis. I can think of the following interactions, and I'd like to get a sense of other possibilities:

 

- Overdriving or other clipping behavior - the guitar example from above could be used by driving many voices through one filter or waveshaper.

 

- FM - if voices could fm themselves polyphonically (sum & difference), something similar to resonant string behavior could be modelled.

 

- Logic - I assume that in a highly complex synth one could program slightly different characteristics based on how many concurrenct notes are being played (or a similar logical test)

 

- Polyphonic filters, (it would be neat if filters behaved polyphically in parallel, where the entire signal (all notes) would pass through each filter and then be summed. This would enable the creation of varying resonant peaks in the timbre depending on which cluster of notes was being played.

 

Please share any experiences or ideas you might have for polyphonic synthesis. Also, any gear... I believe the nord modular and reaktor, will do some of this. I am not sure whether any other synths will.

 

Thanks in advance,

 

Jerry

[Magpel]

Jerry, I often wonder about that myself, and while I'm far to naive to add any insight to the discussion, I will say that I thought this was part of the premise of physical modeling--to implement some of the rules of how instruments behave polyphonically. Am I wrong about this?

 

John

[Botch.]

All the notes you play on you "multi-monophonic" synth will all shake the speaker cone at the same time, and hence can't help but interact. Even if you route each note to a separate out/separate speaker, they will converge and shake your eardrum and interact there. By this definition, ALL synths are polyphonic!

"Eardrum Interaction" may be a good name for a rock band.

[RABid]

Where did you find this definition? Polyphonic literally means multi-note or multi-sound. Long ago I heard the definition you listed applied to the term cross timbre poly-synthesis. Yep. Long name, but descriptive.

 

One trick. Try splitting your signal in a good software modular synth. Send one side (split 2) through some effects which include heavy reverb. Use the effects only part of that signal to modify either the main signal or split 1.

[dansouth]

If you run the output of the "oscillators" through the effects section of a sample playback synth, you can get the distortion effect that you mentioned happens with a live guitar. I've observed this on my Triton when playing distored guitar sounds.

 

I've also noticed that if you play two identical MIDI notes simultaneously from a sequencer on the same synth (something you cannot do playing live from the keyboard unless you layer the same sound over itself), it results in a nasty sounding hybrid, not a natural layer. If you try to double a violin or trumpet patch with the SAME patch on another MIDI channel, it does NOT sound like two violins or two trumpets. It sounds like a very unhappy synthesizer. Synthesized orchestrations have to be adjusted to compensate for this effect.

[aeon]

Jerry:

 

Of the synths I currently own, I know the Access Virus b can do a tiny bit of what you describe, I know the Waldorf Q can do quite a bit more, but as you seem to already know, the Clavia Nord Modular can do all manner of the things you describe.

 

By using logic modules, note detectors, parallel processing paths, per-voice and per-patch signal sections and conditional feedback loops in a Nord Mod patch, you can make the number and voicing of the notes you play have direct control over all manner of parameters and morph groups in the sound. The simple fact that the modulation sources and destinations are so open-ended make the Nord Mod one of the (potentially) most expressive synthesizers ever.

 

Im finally going to get a wind controller to go with it, as I can actually play reeds (but not keys). It is the very qualities you describe along with the open-ended patching that make it so suitable for a wind controller. Combined with the modulation options in my t.c. FireworX, I should be able to program some truly expressive and organic timbres.

[CoolColJ]

Piece of cake on an androemda!

[DougP]

Originally posted by CoolColJ:

Piece of cake on an androemda!

 

Actually, no, not for what the original poster is asking for (the only exceptions being "polyphonic" modulation sources, such as aftertouch, number of keys on, master clock/sequencer, etc. - all of which would also apply to the like of the Triton/XP80/etc.), since components of one voice (be it the VCO, VCF, EG's, or any other single-voice modulation or audio sources) cannot affect any of the other 15 voices. But for "regular" polyphonic (i.e. not "cross timbre poly-synthesis") synths, I don't really see the point. That's why I own an Andromeda AND modulars. If I need "cross timbre poly-synthesis", I'll use the latter ...

[Tusker]

Originally posted by Magpel:

I will say that I thought this was part of the premise of physical modeling--to implement some of the rules of how instruments behave polyphonically. Am I wrong about this?

 

John:

 

When I play some of the modelled instruments, I really want pieces of them in a general purpose synth.. to use in a different context. For example, I would like to use the overdrive features in the Korg CX3 or the B4 on a regular synth voice, because they respond so sweetly to added polyphony. The only other "polyphonic" models that come to mind are the models in the Korg Z1 and the Yamaha EX5. It doesn't seem like any of the others really focus on 'polyphonic' treatments.... yet.

 

Regards

 

Jerry

[Tusker]

Originally posted by Rabid:

Where did you find this definition? Polyphonic literally means multi-note or multi-sound.

 

Well I made it up. It is most inappropriate, and I don't mean to go against twenty five years of synthesis history in this thread. What I wanted to point out was that we missed a turning 25 years ago (for good reasons such as economics) that we could do well to revisit today, when we have the ability to have synthesis resources pre-mix AND post-mix. By using the definition the way I have, I am calling attention to the fact that bundling the components of 8 monophonic synthesizers into a single multi-voice package, does not yield anything more interesting than 8 monophonic synthesizers played simulatneously.

 

I understand that as manufacturers were looking to build multi-note synthesizers, some early attempts were to have a single filter/amp section for ALL NOTES. Some of these were called paraphonic. People held out that this was not as attractive as "full" synthesis features for every note. So the standard that evolved was to treat each voice as individual (i.e. complete and isolated) within polyphonic synthesizers. In fact the much derided polymoog has a (bypassable) single filter through which all voices could pass. This created unusual performance requirements and although the polymoog IS a polyphonic in the way it's presets work, people focused on the single filter as a weakness.

 

As a result of concentrating the synthesis resources pre-mix, we ended up less interesting (but predictable) sounds than if we had concentrated more resources post-mix.

 

I think guitarists evolved a long a more musical path, because they worked with the limitations better. They took (in some cases) pieces of modular synths, converted them to cheap stomp-boxes, and used them, whether one-string or six strings were sounding. An overdrive/distortion which may be extremely unmusical with 6 notes, can very useful and dramatic with 2 or three notes of the right frequency. The same holds for things like ring modulaters and modulated filters, which tend to sound very tame in most of (what I am calling in this thread) multi-monophonic synthesizers.

 

Anyway, that's a rather long-winded explanation for my choice of words.

 

Regards,

 

Jerry

 

[ 12-11-2001: Message edited by: Tusker ]

[Tusker]

Originally posted by CoolColJ:

Piece of cake on an androemda!

 

What applications do you have in mind, colin?

 

Jerry

[RABid]

Originally posted by Tusker:

 

I understand that as manufacturers were looking to build multi-note synthesizers, some early attempts were to have a single filter/amp section for ALL NOTES. Some of these were called paraphonic.

 

I had an Arp Omni II with this feature around 20 years ago. Back then I actually used the synth voice to cover some basic rock and roll piano. The crowd was usually drunk enough not to notice. Imagine going on stage with a 4 octave, non touch sensative, single oscillator synth of this type and playing piano today.

 

It sounds like you would really enjoy either a Nord Modular or Reaktor. Sound design is an art form of its on, and very enjoyable. You will find that many times accidents make the best sounds.

[aeon]

Originally posted by Tusker:

Anyway, that's a rather long-winded explanation for my choice of words.

 

Long-winded or no, it made me think about the issue further, and I understand it even better.

 

The Access Virus b can do a lot of what you describe because of its ability to use aux busses. The output of a multi-voice channel can be passed back into another channel as if that output were the new channels oscs, so many-voice into one-filter (or any other part of the synth engine) can be realized.

 

The Virus b features per-voice and post-voice(s) saturation/distortion within a single channel, so some of the guitar stomp box tricks can be achieved without using an aux. Of course, params for the saturators and aux bus feeds are targets in the modulation matrix.

[tenthplanet]

Wouldn't it just be easier to overdub with a mono synth line. That way you get the interaction as a regular orchestra would, a grouping of individual instruments.

Michael

[RABid]

Overdubbing a mono line is just the opposite of what he is talking about. The vibration of piano strings are affected by other notes being played on a piano. This goes past the roar you get when holding the sustain pedal down. Many people believe it affects the tone. That the notes of a cord on a piano blend a bit differently because of this effect. It is a very interesting concept, and very difficult to program because synthesizers are not designed to replicate the effects that notes of natural instruments have on each other. The change in overtones of a guitar when multiple notes change the distortion amount is the most basic of examples. This is more of an effect of total volume than of which notes are played. Because there are no documented basics to build upon that leaves the imaginative programmers like Jerry to push into areas where few have been before. And I will be honest, those that do arent in any hurry to give up their secrets.

[tenthplanet]

I should have been clearer; the mono-synth idea only works for non-polyphonic instruments. The piano is almost an orchestra in it's own right and as we have seen, harder than all get out to emulate.

Michael

[Tusker]

Rabid - You are very kind. I don't know about imaginative... dissatisfied programmer maybe. I'll try out your post reverb trick. Sounds kinda like a tomita thing. I am happy to share what little I know. Still learning.

 

Aeon - I downloaded the TC's information. That thing is really like a modular synthesizer (effects) isn't it? Mates perfectly with the nord I bet. I hope Clavia comes up with a a digital out for the nord. I'd love to hear anything you come up with on that combo, especially with the wind controller.

 

Colin & Doug - as to the Andromeda... I know the Andromeda allows you to route a single voice into another voice pre-filter. (So as to get up to four filters in series.) I don't know if it allows you to route a handful (say 15) of voices into the downstream filter(s). If it did, I'd guess it has some possibilities, at least in the traditional overdriven-filter/amp kinda of way. (Good to hear the virus does that too.)

 

I was hoping for a some ideas that are less volume (overdrive) oriented (more notes = more effect). I guess ideally one needs a synth that tracks # of notes played, and allows you to build conditional logic using midi note number (so as to use distance between notes as a modulation parameter).

 

That would really open up the harmony & voicing possibilities: stacked fourths could yield one set of overtones and stacked thirds could yield a different set.

 

Thx,

 

Jerry

[aeon]

Originally posted by Tusker:

I downloaded the TC's information. That thing is really like a modular synthesizer (effects) isn't it? Mates perfectly with the nord I bet.

 

Yes on both points.

 

Originally posted by Tusker:

I was hoping for a some ideas that are less volume (overdrive) oriented (more notes = more effect). I guess ideally one needs a synth that tracks # of notes played, and allows you to build conditional logic using midi note number (so as to use distance between notes as a modulation parameter).

 

The Waldorf Q has voice count as a mod source, so indeed the number of notes played can be routed to many parameters or to the function generators. Sadly, the Q docs are not clear on this, so I am waiting on information from Waldorf while I patch.

[CoolColJ]

Well on the Andromeda there is a mod source called Number of Notes held.

 

Just wire that to your destinations and bingo - timbre that changes according to how many notes held

Skys the limit here when combined with the idea of this modulating something which then modulates something and so on. Also Key number could be used. then things can be routed to the tracking generator and modified as well.

 

 

I'll make some mp3s to illustrate this.

 

[ 12-14-2001: Message edited by: CoolColJ ]

[Brittanylips]

This is a SILLY question because:

 

your definition of polyphonic is actually achieved by many synthesizers in many ways all the time with the caveat that the interaction is not a result of playing simultaneous keys -- for good reason. The keys are the interface. If you allow sympathetic programming to alter the precision of the interface, you loose control. Better to have all of that interplay under the hood, available, and then be able to control it. Envelopes, filters, sound sources, etc., frequently interact -- but not to the extent that the interface become ambiguous.

 

If you played two notes on the keyboard and got an unintended third, you take away your ability to make choices and be the musical mind. Better to have the sound available under the hood, and press the note that you know will give it to you.

 

As for modelling symphathetic acoustic phenomena, such as you get on the piano, everything I've seen so far seems real gimmicky.

 

-Peace, Love and Brittanylips

[RABid]

Originally posted by Brittanylips:

If you allow sympathetic programming to alter the precision of the interface, you loose control. Better to have all of that interplay under the hood, available, and then be able to control it. Envelopes, filters, sound sources, etc., frequently interact -- but not to the extent that the interface become ambiguous.

 

 

 

That is one of the major difficulties in programming and one reason I gave the example I did. Note I split the signal, used reverb only from the secondary split, and applied it to modify the primary split. I purposely did not use a sound to effect its self because it is easy to cascade into chaos. This is NOT an area for a beginner. However, if you can achieve that delicate balance wonderful things can happen.

[coren]

i get something like this on the keyboard im borrowing from my parents, it's an old yamaha, casio-like in quality

none of the patchs sound even close to what there called, but thats not important

 

when using something like a trombone, if i hold a key like a low c (i use lower ones cause the following effect is slower..) it sounds normal, holding a chord is semi-normal, but if i hold a note (like low c) then hold the key directly next to it (in this case half tone up) it dosnt sound like the original signal.. it begins a throbbing pulsing sound coupled with other noises and remnants of the original signal... sounds really cool, id use it to represent a power generator or huge engine if i needed to fake it for some reason... it works with several other patchs as well, like the sine wave. actaully, if i use both my hands and press as many low keys at one time that i can, it dosnt even make a tone (one youd expect).. but i can actually feel a physical rumble in the keyboard... and hear some growling.

 

im not sure if its because of the singal speaker or if they just blend it all inside cause its cheap but the keyboard

bounces the notes of each other better than my guitar does

[Tusker]

Originally posted by Rabid:

That is one of the major difficulties in programming and one reason I gave the example I did.

 

I agree that is certainly a concern in this area, because it is so much like FM programming: A little change can make a big difference in result. However that need not terrify us. Three ways to 'control' the outcome are by limiting the effect itself, limiting the technique to a certain layer-slice (as in Rabid's example), or limiting the technique to a time-slice of the sound such as the attack.

 

Perhaps a broader perspective on this trade-off between expression and control is that synthesizers are arguably the most controlled instruments on the face of the earth. We have the ability to define virtually any aspect of a sound within the limits of the technique employed. Given where we currently lie on the continuum between expression and control, I'd be willing to sacrifice some of that awesome control for more expression from time to time.

 

The wonderful, unique thing about the synthesizer as an instrument is that we have the freedom to make such choices.

 

Regards,

 

Jerry

 

[ 12-18-2001: Message edited by: Tusker ]