Randomize pan

[ppadru1]

I want to play a little popcorn sounding staccato lead, but want to randomize the pan on each note.  Is thus possible with the rev 2?

[guyaguy]

I can't think of a way to completely randomize it per note but you could use the gated sequencer in Key Step mode with the destination set to Pan Spread and each step set a different value. It will probably sound random enough, especially if you set the sequencer to reset at the 15th step so it doesn't cycle through 16 steps on a melody in 4/4 time.

[OneLittleFonzie]

I want to play a little popcorn sounding staccato lead, but want to randomize the pan on each note.  Is thus possible with the rev 2?

isn't this what the "Pan Spread" is for?

[jok3r]

I don't have my unit in front of me now, but out of my head I would suggest to set pan mode to fixed in the misc parameters and then modulate the value per mod matrix with a random LFO. I think I did that on some patches my self...

Pan spread does not randomly assign voices to left and right. It is just alternating between the two channels (LRLRLRLR...). Someone should correct me, if I'm wrong...

[OneLittleFonzie]

Pan spread does not randomly assign voices to left and right. It is just alternating between the two channels (LRLRLRLR...). Someone should correct me, if I'm wrong...

you are right!! Ignore my suggestion.

[OakBloodThree]

Pan spread does not randomly assign voices to left and right. It is just alternating between the two channels (LRLRLRLR...). Someone should correct me, if I'm wrong...

I was under the impression that PAN SPREAD set the maximum level of panning, and that each voice would be panned away from center by a different amount, with the additional constraint of alternating direction.  The manual is not completely unambiguous:

Pan Spread: 0...127—Pans the audio in the stereo feld individually per voice. With a pan spread setting of zero, all voices are panned in the middle (monophonic). As you turn up PAN SPREAD, the audio in each voice is gradually moved away from the center by different amounts. Every other voice goes in a different direction. This provides a wide stereo feld as you play. Any modulation to Pan will individually move each voice from its static position as set by the PAN SPREAD setting.

I'd have to inspect a stereo waveform to see...

If you don't have a sequencer channel available, I think you could also get randomness by modulating pan with a very slow LFO set to random waveform and key sync. (I think each keypress generates a new random value).  Of course, it will eventually change to a new value if you sustain the note long enough.

[creativespiral]

Guyaguy's suggestions of using the Gated Sequencer in Key Step mode is a good one.   Just make sure the pan mode is set to "Fixed".

There is one other way you could try...  set a LFO as Random/Noise shape, slowest rate possible, with Key Sync ON.   This should restart the LFO at a different random position each time you strike a new note, and the rate of change on the LFO will be slow enough that you won't ever hear the transition, unless holding notes for over five seconds.   (set layer pan mode to "Fixed" here too.   

[guyaguy]

Guyaguy's suggestions of using the Gated Sequencer in Key Step mode is a good one.   Just make sure the pan mode is set to "Fixed".

There is one other way you could try...  set a LFO as Random/Noise shape, slowest rate possible, with Key Sync ON.   This should restart the LFO at a different random position each time you strike a new note, and the rate of change on the LFO will be slow enough that you won't ever hear the transition, unless holding notes for over five seconds.   (set layer pan mode to "Fixed" here too.

Ah there you go! I couldn’t think of how to connect the key stroke to an LFO but that should do the trick!

[panic]

Guyaguy's suggestions of using the Gated Sequencer in Key Step mode is a good one.   Just make sure the pan mode is set to "Fixed".

There is one other way you could try...  set a LFO as Random/Noise shape, slowest rate possible, with Key Sync ON.   This should restart the LFO at a different random position each time you strike a new note, and the rate of change on the LFO will be slow enough that you won't ever hear the transition, unless holding notes for over five seconds.   (set layer pan mode to "Fixed" here too.

Ah there you go! I couldn’t think of how to connect the key stroke to an LFO but that should do the trick!

Does this actually work on rev2? (I don't have one to try it out). I remember with the Prophet 08 (rev1) somebody reported that the random LFO was actually "semi-random". Meaning that it spits out a series of values that at first sight seem to be random, but it is always the same seemingly random series . Also meaning that if you slow it down, and use keysync, it will always be at the same value on the keypress.

[maxter]

Guyaguy's suggestions of using the Gated Sequencer in Key Step mode is a good one.   Just make sure the pan mode is set to "Fixed".

There is one other way you could try...  set a LFO as Random/Noise shape, slowest rate possible, with Key Sync ON.   This should restart the LFO at a different random position each time you strike a new note, and the rate of change on the LFO will be slow enough that you won't ever hear the transition, unless holding notes for over five seconds.   (set layer pan mode to "Fixed" here too.

Ah there you go! I couldn’t think of how to connect the key stroke to an LFO but that should do the trick!

Does this actually work on rev2? (I don't have one to try it out). I remember with the Prophet 08 (rev1) somebody reported that the random LFO was actually "semi-random". Meaning that it spits out a series of values that at first sight seem to be random, but it is always the same seemingly random series . Also meaning that if you slow it down, and use keysync, it will always be at the same value on the keypress.

Yeah, with LFO rate at 0-20, key-sync and "random" wave, it steps between 8 repeating values. With higher/faster rates, the values change from time to time.

[guyaguy]

Guyaguy's suggestions of using the Gated Sequencer in Key Step mode is a good one.   Just make sure the pan mode is set to "Fixed".

There is one other way you could try...  set a LFO as Random/Noise shape, slowest rate possible, with Key Sync ON.   This should restart the LFO at a different random position each time you strike a new note, and the rate of change on the LFO will be slow enough that you won't ever hear the transition, unless holding notes for over five seconds.   (set layer pan mode to "Fixed" here too.

Ah there you go! I couldn’t think of how to connect the key stroke to an LFO but that should do the trick!

Does this actually work on rev2? (I don't have one to try it out). I remember with the Prophet 08 (rev1) somebody reported that the random LFO was actually "semi-random". Meaning that it spits out a series of values that at first sight seem to be random, but it is always the same seemingly random series . Also meaning that if you slow it down, and use keysync, it will always be at the same value on the keypress.

Seems pretty random to me. I got something like this at Amount 100% for a hard right ot left pan:

rlllrrlrlrlrrlrlrlllrrllrrllllllrllrr

[maxter]

As each voice has its own LFOs, when rate is set low enough with key-sync on, the initial value doesn't change. Which in my case, an 8-voice Rev2, gives 8 repeating values, 1 for each voice. Using faster rates, the values are allowed to change, but then you may also get the S/H jumps while the note is playing. So the LFOs aren't optimal for random values at note onset.

I'd say the best way to go for a semi-random panning would be to set up a sequencer or two routed to pan.

[Razmo]

To expand on the gated sequencer, and asuming you want a new random pan value on each keypress, try this:

Set the gated seq play mode to stepped.
Write in random numbers between 0 and max on all four sequences.

Set up four mod matrix slots with each having their source set to sequencer 1-4. Set the destiantions to pan, and their amounts to 32 (effectively letting each sequence to sum to pan at one fourth per sequence)

Now set the sequence length of each track to a different value so that you do not just get 16 predictable panning values... Eventually let track 1 be 16 steps, track 2 15 steps etc... This way you would get new random values for a very long time without noticing when the pattern gets repeated again.

I have not tried this though, but in theory it should work.

And do not forget to set pan mode to fixed.

You could use fewer tracks if you just scale the mod matrix amounts, which would be handy if you want to use a track for something else, but the less tracks the shorter the repeating pattern will get of course...

Also, you may have to create a mod matrix slot more that has DC as source, and destination PAN, with some negative amount to center the modulation properly... A sequencer value of 0would most likely appear at dead center, so the DC may be needed to shift the panning so that the modulation range is from - 64 to +64...I believe that the sequencer values are unipolar, but the pan destination is bipolar.

[maxter]

It does indeed work as you suggest, Razmo.

I have mainly used this technique to approximate poly-rhythmic quasi-"generative" melodies (stacking both layers) and "random" detuning of OSC's. It's probably the closest thing to a "random" onset (per voice) on the Rev2, though it's actually anything but random. 

[jok3r]

Perhaps it could be good to choose prime numbers for sequence length... I don‘t have the exact mathematical laws in my head at the moment... it‘s just an intuition, of which I think it could be worth a try.

[Razmo]

Perhaps it could be good to choose prime numbers for sequence length... I don‘t have the exact mathematical laws in my head at the moment... it‘s just an intuition, of which I think it could be worth a try.

May certainly be the case... Not to mention what numbers in the four sequences would gkve the most even "randomness"... Lots of experimentation possibilities here 

[maxter]

Ok, sorry for drifting a bit off topic here.

Prime numbers are nice. Although I like to maximize the number of runs before any repeat. So instead of 5 steps, 15, a multiple of 5. Instead of 7, 14 steps. Gives longer "phrases", less obviously 5 or 7 (with 15 I could phrase it distinctly as a 7+8,) and less repeating. 5x7=35 runs, 14x15=210 runs.

I ususally use the highest step count for sequencer 1 of each layer, when making poly-rhythmic stuff, because of the step "rests" which makes for rhythms.

Something simple like this, to start with:

Layer1
Seq1 - 15 steps
Seq2 - 14
Seq3 - 13
Seq4 - 12

Layer2
Seq1 - 14 steps
Seq2 - 13
Seq3 - 12
Seq4 - 11

Seq1 - Osc1 pitch
Seq2 - Osc1 pitch
Seq3 - Osc2 pitch
Seq4 - Osc2 pitch

If I want something straight forward to keep the "beat" of 4/4 (16 steps) I use something outside of the Rev2, but if the sequencers are set up well there's usually no need for it.

I route the sequencers to other stuff as well, in the mod matrix, sometimes even a sequencer with negative amount to the "other" OSC, for example Seq2 - Osc2 pitch- (as it's already routed to Osc1 pitch+) = contrary motion, but because of the amount of other pitch modulations going on it may not be perceived as such, at least not every run.

As for what numbers to put in... Depends on what you're using it for, but trial and error is inevitable for the best results anyhow, so just go at it.  And therefore an EDITOR is key for these things, being able to tweak all 8 sequencers, 8 LFOs and the mod matrix simultaneously as you go along (ie the stuff more or less "hidden" on the front panel).

This method could of course be used with PAN as destination as well. Maybe Seq1 and Seq2 to PAN directly, and Seq3 to control the amount of an LFO which in turn modulates PAN, for some movement (on some of the notes) in the stereo field.


More complex modulations take more effort, but are also more rewarding!

[OakBloodThree]

I thought I had posted this on another thread, but since it's come up...

Prime numbers are nice, but what's important for maximizing sequence length before repeat is coprime numbers, i.e. numbers with no prime factors in common.  (Or at least maximizing the number of prime factors.)

If you do 16 (2×2×2×2) against 15 (3×5), they have no factors in common, so it repeats every 16×15=240 steps.

If you do 16 against 14 (2×7), they share a factor of 2, so it repeats every 2×2×2×2×7 = 112 steps.

If you do 16 against 13 (which is prime), you get 16×13 = 208 steps.

With 4 sequencer tracks of maximum 16 steps, the best you can do is 16×15×13×11 = 34,320.

11 is prime and since 12 = 3×2×2, it shares all of its prime factors with 16 or 15 and wouldn't extend the run length at all if we used it instead of 11.  You would have 2×2×2×2×3×5×13 = 3,120.

[jok3r]

OakBloodThree was faster than me. I wanted to write basically the same, because I was thinking about that the morning.

The mathematical background can be read in this wikipedia article:

https://en.wikipedia.org/wiki/Least_common_multiple

If you really want to max out the chaos and want to use layer two, you should pick the numbers 16 13 11 9 for layer 1 and 7 5 for layer two. Even if you use more numbers for layer 2, you can not achieve a longer sequence, because all possible prime factors are already covered by this numbers. The value you get will be 720,720.

EDIT: At the moment of writing I realized that you can not route mod sources from layer two to layer one, to max out the sequence of the same mod destination. So above numbers are only correct if you count combinations of mod destinations...

[maxter]

Good points, on maximum "randomness" to a mod parameter such as pan.

OT again:
In the case of poly-rhythmic evolving melodies, it's somewhat different because of more than one mod destination (the OSCs). In that case I think of the sequencers as pairs (1+2,3+4) to each oscillator (generally), and find it better to keep the sequencer pairs 1 step apart.