Deeply impressed with the Endless Processor, and definitely agree that it comes alive with external cv and gates. I’ve been working with using a basic single oscillator voice played into two channels of Dual Looping Delay, thru Dual Ring Modulator and into Endless Processor. Below are two takes on this with random control of Endless Processor memory, channel, and capture from Marbles.
What I find most interesting, which you can hear most in Antares, is how the EP can effect the sound. Those noise and metallic sounds are all just a few harmonious tones looped thru a frequency shifter, normally you’d expect that vaguely ring-modulator-like sound. Instead it creates something more complex and organic.
We updated the manual to reflect a hardware change in the most recent batches.
If you will receive (or already received) the Endless Processor with revision 1.3 (see the back of the module, it is written on the PCB), Input 1 is normalized to Input 2, which is more convenient than mult or stack-cable:
Andrii, When you have a chance, I wanted to ask a few questions about how the Endless Processor handles gates received at the analyze input.
I know the EP can only analyze one signal at a time, and I know it can’t cache requests and perform them sequentially (wishful thinking), but I’m curious what the EP does when it receives new gates, along with a changed layer assignment, before an ongoing analysis is complete. Does it “drop” the previous analysis, or does it not allow a new one until the previous analysis is complete? What impact do the fade in/fade out settings have on this?
I ask as I’ve noticed in using the EP with randomized control over layers and analysis activation, that sometimes less new layers are activated than I’d expect from my clock rate. Any suggestions you could share to best use the EP with this sort of external control would be welcome.
Does it “drop” the previous analysis, or does it not allow a new one until the previous analysis is complete?
So, if there is ongoing analysis, the next one will be performed right after the previous one is finished.
(only after the analysis will be finished you will have the ability to launch a new one).
sometimes less new layers are activated than I’d expect from my clock rate.
As the processing of a new buffer involves heavy computation, it is not lightning-fast.
So even if you’re sending fast gates, only when the next calculation is unblocked it will be taken to account and perform the next processing.
The currently selected layer and the channel, fade-in, and fade-out will be taken into consideration - if a layer index is modulated fastly via CV and a new trigger for calculation is received, it will use the layer index at the moment of that trigger.
What impact do the fade in/fade out settings have on this?
Positions of those settings are taken when you’re launching a new process.
When the calculation is started, turning those knobs will not affect anything for this layer.
You can choose slow fade-in/fade-out, process layer 1, turn those to be faster, and process layer 2.
Layer 2 will be faded faster, while layer 1 will keep fading, uninterrupted.
Meanwhile, I keep trying to make calculations of processing to be faster.
In case of success fast triggers will then react more promptly.
Thanks, that is really helpful. Basically the EP will not truncate an in-process analysis, and will only accept gates when that process is complete. Regarding the fade-in/fade-out times, I understand they are only polled when starting an analysis, and aren’t continuously active. What I was curious about is if the fade times add to that length of the analysis, or is the EP ready to do a new analysis before the fade out/in is complete? Minor points, just want to have a better understanding of how the EP works. Very pleased to hear you’re looking at fine-tuning for the software, but I have to have it’s pretty near perfect right now from my perspective.
What I was curious about is if the fade times add to that length of the analysis, or is the EP ready to do a new analysis before the fade out/in is complete?
Suppose your fade-in time is long and started to fade and then you click to re-process somewhere in the middle. It will calculate a new buffer at the point you clicked even if it has not finished fading. So it will begin to fade out and then will fade in with a newly calculated buffer.
Really inspiring stuff! Is there a forecast date for the next batch?
Here’s my first Jam, the entire thing was about 30 mins, but during the last 12 mins I really got in the zone.
This was fun, despite using the module at its most basic (no CV modulation), it’s very playable and I really can’t wait to spend some more time with it.
Sounds terrific! May I ask what that little 2HP module is between the Mimetic and the SSF module?
It looks like the Bubblesound Booster. Great simple utility for boosting line level signals to modular levels.
The demo does sound great. I’d really like to get my hands on one next batch.