unusual noise generators

[martinvicanek]

Here is an illustration of two uncorrelated random sequences:

UncorrelatedRandomSeq.png (31.26 KiB) Viewed 806 times

I took two instances of the popular portable quick and dirty whitish pseudo random number generator with different seeds. "Pseudo" means that the random sequence is in fact strictly deterministic and will be the same each time it is triggered. With different seeds (= starting points) you will get different sequences, even though the algorithm is otherwise identical. Good random number generators will produce white noise no matter what seed you choose (the one in the image may not be as good).

So once you have two different random sequences with the same spectral characteristics you can lowpass them etc.

[tester]

I see. Okay, I will check that when I'm on my SM computer, thanks.

//edit:
Just checked. Works fine. Thanks for additional explanation on that.

[tester]

Here is another small thing. I added a switch to select between common and different seed for left/right channel noise. Common seed will produce mono sound, different seeds will make independent channels. Works fine when turning on and off the soundcard. Now - how to reset the sound generation on seed switch, without turing off/on the soundcard? Like hard resync node.

[oddson]

'Clear Audio' primitive will do exactly this -- restarts all running audio processes.

[tester]

Okay, thanks for advice!

[oddson]

You may prefer to use an external seed instead (see my attachment above) as Clear Audio will disrupt everything and not just restart the noise generators.

[tester]

How the external seed will change anything? What I need to reset on seed change (from different per channel to common for both) - is the state of these noise gens. Otherwise - even if I change the seed during playback - left and right channel will not synchronize (no mono sounding output so to speak). The concept is to switch between random patterns and one pattern that can be spatialized manually via L/R delay. Am I missing something?

[Chaoslala]

I like to use it like this !:

"streamin in;
streamout out;
float rand;
stage(0){rand =0.1001;}
stage(2)
{ rand =in;
rand = rand*3;
rand = rand - (rand>2)&1;
rand = rand - (rand>1)&1;
out = (rand*2 -1)/20;
}"
strangly it preserve the note of the input and make a little overdrive.
Be carefull with the output volume if you try !

[oddson]

How the external seed will change anything? What I need to reset on seed change (from different per channel to common for both) - is the state of these noise gens. Otherwise - even if I change the seed during playback - left and right channel will not synchronize (no mono sounding output so to speak). The concept is to switch between random patterns and one pattern that can be spatialized manually via L/R delay. Am I missing something?

No... I was missing something. I was thinking about de-syncing only and not re-syncing.

And my reference to seeding was for a schematic I never posted...

[martinvicanek]

This is how I would implement an external seed:

RNGwithExternalSeed.png (20.63 KiB) Viewed 596 times

[tester]

Meanwhile I have discovered something else. It's a conceptual randomization for LFO based on trigger-shooter with delay and dezippers (not finished yet), but is fast, provides somewhat good sound values (see the FM example) and at least I know which parameters are under my control . Also - I can put the stream of greens through some "primitive" calculations that I used before (like linear to log random spread), so there are various directions to check.

[tester]

So here is my random stereo generator for FM and AM applications (schematic shows both example settings).

It is fast, made on primitives, various modifications can be added without SM coding knowledge (like random spread conversion/type). One thing is "cheated" here a little bit, but it should not be a problem. "Speed". While input values are in Hz, output is in miliseconds on a integer basis (both, for counter and dezipper). The higher "speed" value, the greater frequency step. But - it is for slow speeds, so this inaccuracy is not so important. Besides - this is random LFO. "Speed" represents how often generator should change it's value, but - vaules to change are in random range and they are passing through dezipper to become a signal, thus - it is still randomness, although triggered on a regular basis. The fact that generators have common basis - may serve for better output - more pleasant, because this is what it is about - pleasant randomness/irregularity. And this one is random and not just irregular (at least if the SM random primitive uses some random basis).

I resigned from "common seed", because after playing with delay as an equivalend of "phase offset" - I came to the conclusion that this is not needed for my little project.