Jump to content


Please note: You can easily log in to MPN using your Facebook account!

Random thoughts on synthesis


Tusker

Recommended Posts

How much randomness should you program in your sounds? How do you program differently for polyphonic sounds than for monophonic ones?

 

These are two questions that are occupying me this morning and I thought I would share them with you.

 

Yesterday, I find myself driving back to Dallas from Austin (3 hours) and I pop in the ELP double live album (King Biscuit Flower Hour) to shorten the trip. And immediately of course I am reminded of the recent threads regarding the CS-80. Because the star of this show is not the modular but the GX-1 (I assume). It's big and round and just as dangerous sounding as the big Moog. And yes my trip went very quickly. "Pirates" particularly was the song where I could contrast the GX-1 performances with a "Live at the Albert Hall" rendition where Emerson used romplers for the orchestration. The difference is night and day to me. I'll take the GX-1's artificial sounding (but dangerous) orchestration over the romplers any day.

 

But it got me thinking about how, we typically see sounds becoming less present when we move them from monophonic to polyphonic status. Yet this GX-1 sounded like the polyphony made it larger rather than smaller. (Not that it was any slouch in monophonic lines.)

 

So I start thinking about how I would program for "polyphonic bigness." The goal would be to make every note sound different. This morning I power up my trusty AN1X, quickly sketch a blippy synth-horn-stab and get to expermenting...

 

1) Ok, in most plucked envelopes, higher notes ring shorter, so let me modulate filter decay inversely with midi note. Ah yes, it opens up that sound, and it begins to breathe. Try that with attacks and it really comes to life. I'm beggining to enjoy sounds with really fast (clicky) envelopes at one end of the keyboard and slow brassy swells on the other. These are sounds I could play for more than 3 minutes without apology.

 

2) Ok, I want more control, so let me modulate attack and decay times with velocity as well. Defintely a difference.

 

3) But what about randomness? Ok let's try the same routings with random modulation rather than velocity and midi note. The result, even more dangerous. (Disclaimer: I'm not aspiring to make my lowly AN1X sound like a CS80 or GX1. Just trying to make it sound the best it can.)

 

But here is my philosophical quandary. There is some performer's pride that gets in the way. Even though the results aren't as cool, I prefer being in control of the modulation (velocity and midi note) rather than let pure randomness take over. Even though I understand that "natural instruments" like pianos do have randomness in them and I don't assert my "performers pride" asking for perfect sameness in every note when I play them.

 

Do you run into this? How much randomness do you program and how? Do you let chance have a chance in your sounds?

 

Best,

 

Jerry

Link to comment
Share on other sites



  • Replies 9
  • Created
  • Last Reply

This is an excellent post Jerry. Agreed on the ELP sentiments as well. The ELP rompler years really were a let down for me, but Keith's playing made up for it. ;)

 

You wrote: So I start thinking about how I would program for "polyphonic bigness." The goal would be to make every note sound different.

 

You've raised some interesting points about programming unique patches, such as adding voltage controlled random elements vs. direct human control. I think it's best to use both.

 

Routing things like velocity to things other than just the conventional filter cutoff can help liven up a patch. Try varying osc sync effects, slight osc detuning, LFO speed, or filter resonance via velocity.

 

A very cool thing with the ARP Odyssey (and other synths) is that you can use a foot pedal to control Osc sync effects or filter cutoff. This frees both hands to play as well as add more articulation with pitch bend, vibrato, or front panel tweaking. This can really sound great when you're modifying all these things live while you play.

 

One of my favorite things to do (which you also touched on) is to vary certain parameters across the keyboard. The Oberheims with the tracking generators are great for having things happen in each of the 5 octaves. The whole patch takes on a different feel by applying tracking generators. They are one of my favorite programming tools, but not very many synths have them. Keyboard scaling where you can specify breakpoints (DX7 and others) are the next best bet, imo.

 

Also don't forget that external effects can spice things up. Other times, they tend to mask the pureness, or richness of the sound of some analog synths (again, imo).

 

Just have to try things out, I guess. Experimenting is part of the fun!

Link to comment
Share on other sites

I find the use of random-nature modulation sources to be of critical importance as it concerns many digital instruments.

 

A perfect example is my E-mu XL-7...its base sound character is very precise and exacting, but clever use of the cords (mod matrix) and random sources can really open up the sound.

 

The E-mu engine provides two random values for each new key-on, and these can be routed to damn near anything, including other modulation routings. By using these sources routed to many destinations at once, with both positive and inverted bias, you can get some synth voices that really open up.

 

The other random source that can deliver great results is lag-processed pink noise...lovely when routed to osc pitch. Does this make the osc sound like a VCO? no...of course not!...but it does liven things up in a very organic way. ;)

Go tell someone you love that you love them.
Link to comment
Share on other sites

Random thoughts on random thoughts... :D

 

The mono vs. poly thing is really important. Let's examine mono sounds first.

 

Let's look at what players of traditional instruments instinctively do, a clarinetist for example. When the clarinetist is playing a solo line, the sound is continuosly changing - he can afford to do that, because if he has the 'singing' line, all the attention is on him at that moment. Are those changes (in timbre, loudness, vibrato, etc.) random, at least in part? In my view, the randomness is really minimal. A good instrumentalist, with a good relationship with his instrument, knows what to do to bend the sound to his will, and to 'let go' to the idiosyncracies of the instrument when it's the case. So the randomness is very, very controlled. What makes the sound come alive are, in my opinion, two major factors:

- The sheer beauty of the instrument's sound;

- The above-mentioned control from the player.

 

Now, can a synthetist compete with that kind of expressivity? I think the answer is yes, but only if he is willing to take the time to really learn his controllers and stuff. Let's take vibrato as an example. You can program a beautiful vibrato by slaving LFO's amplitude and/or speed to velocity, aftertouch, wheels, etc., then modulating very slightly the LFO rate with another LFO or other things, so the cycles don't sound too 'electronic' (an old trick of mine). But I think the best way is to give vibrato directly with a physical gesture on a controller, be it a wheel, or, better yet, a ribbon (my favorite).

Same thing for filter or other timbre-shaping parameters: You have to find the controller that works best for applying it, then really practice its technique. For timbral changes, I prefer a pedal, a slider, or a wheel.

 

Now, notice that most of these techniques take two hands. This is totally normal to me; it's like that in all other solo instruments. With winds, you use both hands to choose your note (and support the instrument!), and you use your breath and lip pressure to shape the sound.

With string instruments, the left hand has the choice of note and all pitch informations, while the right deals with intensity and timbre.

 

Let's now examine what our clarinetist do when playing, say, a Bach fugue with three other clarinets (let's say, a piccolo, a Bb, an Eb, and a bass clarinet... :) ). He will play with "solo-style" expressivity, but also will be careful not to go against the overall texture. In other words, he will still play cantabile, melodic phrases, only with attention to the overall dynamics with respect to the other players, and also less small note-to-note differences, which would go unnoticed in this context and would subtract from the overall sound's coherence.

 

Can a synthesizer player duplicate (in real time, of course) the expressivity, interplay and cohesion of the clarinet quartet? In my experience, the answer is no. If he has to deal with all the controllers that I referred to above, he could only manage one line at a time of real melodic playing. Even with polyphonic aftertouch, it would be an hyperhuman task to try to think to multiple lines simultaneously.

Maybe we would need controllers like the Mixtur-Trautonium, which sadly, never went in common use... listen to what Oskar Sala does with polyphonic lines on that instrument. But even him, a virtuoso who played the same instrument for 60 years, can't play more than 2 lines at a time with absolute control.

 

OK, now let's assume our brave and versatile clarinetist is hired to play in an orchestral work, again as part of a four-clarinet group.

Let's say that the composer tends to treat the four clarinets as a chordal ensemble, giving them sustained four-note chords, maybe with included crescendos and diminuendos. The four players will tend to uniform their sound as much as possible, and also to synchronize their attacks and releases, play in tune with each other and with little or no vibrato, etc.

 

Needless to say, for the synthetist this type of playing is a piece of cake. He simply plays the chords, giving the appropriate dynamics to all the notes, and he programs the sound with the appropriate modulations to give an "ensemble" feel, but without introducing too much animation.

 

So, how much randomness should one introduce in his own sounds? The only answer, to me, is, 'it depends on the type of part that you're going to play'. If you play an exposed, 'singing' part, then you can 'move' it quite a bit, then you can introduce an higher level of randomness, *because you're also going to control and change the sound a lot with your gestures*.

Generally, the more a part is buried in the orchestration, the less dynamics it receives from the player (I'm talking about small, note-to-note dynamics here, not general piano and forte), so less randomness is appropriate.

Theoretically speaking, I like to give some kind of random changes to my sounds, but not more than 10/15% of the amount of change I'm reserving for my playing - in short, my own decision.

 

As I said, I like to give a bit of randomness, or at least unpredictabilty, to cyclic modulation of pitch, timbre, and sometimes volume; that means, usually, LFO modulations.

Other candidates are timbre itself (filter, resonance, FM amount, crossfades, sync, ring mod...), envelope segments (for example, amount of pitch envelope attack), asymmetrical response along the keyboard, etc.

 

There remain the matters of 'playabilty' and 'polyphonic sound'.

 

About the first - I'd like to hear more synth players who really try to master their realtime controllers. I know synths can go well beyond the 'quack' sound most people, sadly, identify them with. Let's show that to the world.

 

About the poly vs. mono sound, it's obvious that certain 'fat', larger-than-life sounds, when played polyphonically, can become a sonic mud. After all, one of the reasons those sounds are fat are the ever-shifting harmonics that never move the same way twice. Multiply those harmonics movements by five or six, and you have such an army of moving harmonics, you ears could forget where the fundamental is. :D So in this case, you'll have to reduce the amount of detune in the oscs, or the amount of osc themselves, the amount of modulations, etc.

 

That said, I would prefer to have a choice... so if a real analog, *polyphonic* modular Moog ever appears on Earth during my lifetime, I definitely want it.

 

The moral: Program your sounds according to the tune. :D:D

 

Wow, did I really write all that 'random' stuff? :D

 

Carlo

Link to comment
Share on other sites

Originally posted by marino:

I'd like to hear more synth players who really try to master their realtime controllers. I know synths can go well beyond the 'quack' sound most people, sadly, identify them with. Let's show that to the world.

 

About the poly vs. mono sound, it's obvious that certain 'fat', larger-than-life sounds, when played polyphonically, can become a sonic mud. After all, one of the reasons those sounds are fat are the ever-shifting harmonics that never move the same way twice. Multiply those harmonics movements by five or six, and you have such an army of moving harmonics, you ears could forget where the fundamental is. :D So in this case, you'll have to reduce the amount of detune in the oscs, or the amount of osc themselves, the amount of modulations, etc.

Probably the best example of someone playing a synth polyphonically was Eddie Jobson playing a Yamaha CS-80 on the original UK album. Truly a master performance, from the comping behind "In The Dead Of Night," the sheer speed and complexity of "Presto Vivace," the string pads on "Thirty Years," the power of "Alaska," and the soloing on "Nevermore." As far as I am concerned, no one has really come close since....another thing about this performance is POLYPHONIC AFTERTOUCH. Synth players are severely limited if they can't affect notes individually. I am probably starting to sound like a broken record regarding this, but at this point there is no excuse for lack of PolyAT!!!!!
Link to comment
Share on other sites

What lovely reponses.

 

Odyssian, we have a lot in common when it comes to creating air in the synth. I often look to the oscillator section to create harmonic variation. It may be my lack of exposure to good filters, or the fact that I have been spoilt with romplers, but there's nothing as good as carefully injecting some "random" inharmonic material in the oscillator section via sync or fm or something.

 

Marino, thanks for sharing the three models from the orchestral instrument metaphor. It does depend on the part one plays. And sometimes, one plays something completely different. One key I think (to borrow from the classical metaphor) is to decide between conscious (where the listener is conscious) randomness as a musical statement, and randomness as a tool of richness. The brain can take more randomness if it perceives some kind of artistic intent behind it.

 

Part of the problem is that today's ears are used to precision, a successful synth has to be designed for both ends of the spectrum ... and the "randomized oscillator pitch" just doesn't do it by itself. One bummer about the AN1X synth is the monophonic LFO. You just can't get those "speeding up the LFO toward the upper end of the keyboard" effects that give each polyphonic voice it's own place in the sound.

 

Originally posted by aeon:

The other random source that can deliver great results is lag-processed pink noise...lovely when routed to osc pitch. Does this make the osc sound like a VCO? no...of course not!...but it does liven things up in a very organic way. ;)

Aeon, are you referring to the EMU here. Or is this a nord modular / softsynth technique. I didn't realise the EMU engine had a lag processor or a choice of noise colors. It's nice to have two random number generators instead of one though isn't it? Do you have some favorite destinations for subtle randomness? Mine are typically in the oscillators (pitch, synch pitch, etc.)

 

Marzz, sadly I haven't listened to Jobson much. I'll pick up the album sometime. Two of my favorites albums are Bladerunner and Chariots of Fire and although I can't say it with certainty, I imagine poly-aftertouch gets a workout on those two albums. Have you done some Poly AT applications? DO you have some preferences? I've occasionally considered picking up a Poly AT controller (like an A-50), but wondered if the ensuing MIDI problems would be worth it.

 

One issue with randomness in synths ... I think we focused so much on keeping things controlled, synth design never embedded randomness in ths process. For example, synths don't typically

allow feedback loops for midi, audio and controller information ... they don't allow voices to interact with each other until the EFX section typically ...

 

I imagine that will change in the next few years, as synths become viewed as the general purpose

sound processing computers they are. Perhps we will have another modular age. :thu:

 

Best,

 

Jerry

Link to comment
Share on other sites

Jerry Aiyathurai said:

I often look to the oscillator section to create harmonic variation. It may be my lack of exposure to good filters, or the fact that I have been spoilt with romplers, but there's nothing as good as carefully injecting some "random" inharmonic material in the oscillator section via sync or fm or something.

Or sweeping a wavetable on a MicroWave XTk! :D

 

Jerry Aiyathurai said:

Aeon, are you referring to the EMU here. Or is this a nord modular / softsynth technique. I didn't realise the EMU engine had a lag processor or a choice of noise colors.

This is on the E-mu...and yes, it has not just a lag processor, but all manner of arithmetic/logical operators in the cords (mod matrix) section such that control signals can be shaped and processed.

 

Jerry Aiyathurai said:

It's nice to have two random number generators instead of one though isn't it? Do you have some favorite destinations for subtle randomness? Mine are typically in the oscillators (pitch, synch pitch, etc.)

Yes, it is nice to have two random generators...but that is per layer, so there are 8 random generators per patch, and up to 24 of them in a linked patch!

 

Of course, those are sampled at each key-on, so if you need real-time, continual/changing sources of randomness, use an appropriate LFO shape, or the white/pink noise sources (usually shaped by a lag processor).

 

As it concerns the E-mu, my fave destinations for random control signals are the osc pitch, osc fine pitch, sample start, sample loop (after a gain 4x scaler to add some grain and grunge to the osc), sample retrig (on drumsamples for patterns that are programmed, but whose output varies in a semi-random manner), the attack segment of the many envelopes, LFO speed, the filter cutoff and Q params (especially since many of the filters are EQ shapers and formant things as opposed to typical LP/BP/HP stuff), and perhaps most of all, the scaler value for the modmatrix routings themselves, because by doing this, you can randomize not just the target params in the synth engine, but the degree to which those params are being modulated.

 

Jerry Aiyathurai said:

One issue with randomness in synths ... I think we focused so much on keeping things controlled, synth design never embedded randomness in ths process. For example, synths don't typically

allow feedback loops for midi, audio and controller information ... they don't allow voices to interact with each other until the EFX section typically ...

The Waldorf Q can sense how many voices are being used and make use of that as a mod source...that can be killer when making a patch using the XPhorm function such that low voice count optimizes the patch for monosynth voicing, and high voice count morphs the voice to a polysynth implementation.

 

Of course, the Nord Modular can do all manner of feedback paths and control signal shaping, but I intended to speak more about the fixed-architecture synths I am familiar with.

Go tell someone you love that you love them.
Link to comment
Share on other sites

But here is my philosophical quandary. There is some performer's pride that gets in the way. Even though the results aren't as cool, I prefer being in control of the modulation (velocity and midi note) rather than let pure randomness take over.

 

How much randomness do you program and how? Do you let chance have a chance in your sounds?

 

Synth players are severely limited if they can't affect notes individually.

 

This philosphy held me back from getting into the 'self-evolving' patches I'm experimenting with now. Play a note then let the sound evolve for 10 minutes or better. Get four or five of them going(out of sync of course) then play over the top.

 

Self-evolving sounds harken back to the modular days before keyboards were connected to synths. Currently I'm having filter and amp envelopes, as well as frequency and pulse width modulated by seperate LFOs with different wave forms at different rates to achieve as much randomness as possible. Using analog style sequencers(I'm using VAZ Modular) playing different notes at a slow rate, say 2 or 3 per minute, can add to randomness. I don't think Keith would go in for this. :)

 

Steve

You shouldn't chase after the past or pin your hopes on the future.
Link to comment
Share on other sites

Steve:

 

That is exactly the consideration, I believe I'm missing out on. It seems that with synths, you need to be willing to adjust your mindset as you go.

 

Monophonic Mastery - where you might do the vibrato yourself, disdaining the services of the LFO

 

Polyphonic Programming - where you might pre-program some random effects and perform others.

 

Twisty Noodles - where a self evolving patch may have some guidance from you during the performance, or it may not.

 

There are others I'm sure.

 

Jerry

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...