Thread: Wiard Noisering
View Single Post
Old 11-04-2010, 06:27 PM
baltimoroder baltimoroder is offline
Join Date: Mar 2009
Posts: 110
Default Wiard Noisering

Thought some of you might enjoy this... A Grant Richter/Wiard "Noise Ring"-inspired sequencer stack in Numerology 3 (pretty sure you need PRO PR1 Build 12.5 or higher to run the project.) More info see:

Here's a test recording on soundcloud... I let it run on it's own for a minute, then adjust CHANCE & CHANGE.

Hit play. The Noise Ring will generate random, but _recurring_ pitch motifs over the Gate Seq you specify.

- "Gate Seq" sets the gate sequence. Nothing funny happening here.

- Increasing "CHANGE" makes the sequence more un-stable. Decreasing the fader decreases entropy. At "CHANGE" = 0, the sequence effectively STARTS LOOPING indefinitely.

- Increasing "CHANCE +/-" tends to throw more 1's at the shift register. Decreasing the fader increases the 0's. Too much in either direction, the sequence locks into a single note. Think of this as a rough pitch "guide"; e.g. if the sequence is only playing low notes, bump this up.

- The "CLOCK" does what it says. In the original design, increasing the Ext Rate knob drove entropy into the clock speed. I left this off to make it more useful. Easy enough to wire up if you feel like it.

- "RING X-FADE" fades between n+1 and 2^n outputs. At the lowest setting, you'll only hear a note or two. All the way up is fully 2^n. Since notes are binary, this acts to limit the range of possible notes. The original module has 2 separate outputs.

- "DELAY" simple wet/dry delay.

- "MEMORY SWITCHES" turn individual memory bits on/off. Notes get swapped or reduced out entirely. If CHANGE is set to 0, these are good for generating _consistent_ sequence variations, otherwise it's hard to hear what they are doing. MSB-128 has the greatest effect. Original module defintely did not have this feature.

- Freeze a sequence by turning the "CHANGE" fader all the way down. While frozen, you can mess with the "RING X-FADE" and Memory Switches to get maximum variations on the original 8-bit data. Or sequence these parameters separately.

- Be careful of too much difference between CHANGE & CHANCE. I've found for maximum "musicality", they prefer to be offset from each other around .3. CHANGE seems to be a bit more forgiving with it's setting but sometimes regardless of what you do the sequence will go to sleep and sound like it's stuck, moving both knobs or stop/start usually clears it.

In the original design, white noise is run through 2 different comparators. They generate two (inter-related) streams of 1's and 0's. These two streams drive a Shift Register; one fills it with data (CHANCE), the other stream controls whether entering data is new or "recycled" (CHANGE). The data in the register is 8-bit memory which is then converted to generate pitch information at a rate set by the CLOCK. The outputs n+1 and 2^n are both generated (pretty sure I did this right...)

Attached Files
File Type: zip (35.2 KB, 117 views)

Last edited by baltimoroder; 11-04-2010 at 06:32 PM.
Reply With Quote