Clouds Successor


I’d really like to see a MI end-of-chain fx unit.


The problem is that people expect this kind of module to be a “kitchen sink” module with the whole collection of FX types, and I’m done with that type of thing (consider Plaits my last attempt at this module type)…


Part of the fun would be to see you subverting expectations :smiley:


I think there’s still a lot of space around delay/reverb/flange/chorus/phase effects that could be explored in an interesting way, with a carefully calibrated set of macro controls.


Or just a time-based effect with a very wide range.
Flanger-chorus-delay range is what i mean.
No macro controls, the dials are the same for all effects, type is just based on time settings and external modulation.

Except for the mungo D0 which is very spartanic there are no delay units in euro that nail both the very short and the very long end of the delay time spectrum.

It could fit in 14hp, maybe make it dual and add some cool macros like spacing between delay times or morphing between serial and parallel or feedback routings


Also something i bet would be brilliant would be a room simulation with following parameters:

Room size (from snare drum to factory)
Absorption (from resonating to immediately damping)
Material (wood, metal etc)
Position in the room

There would be an input for reverb as well as one for striking the room like a drum.
When nothing is plugged in, the strike in can be switched to a trigger input for striking the room with an internal exciter.
Self oscillation with absorption all the way ccw tracks v/oct


The Eventide EuroDDL does really really short delays and also very long ones, with customizable ranges for CV input so you can also modulate over either small ranges or large ones.


I agree about the Eventide; It is excellent, and for me, puts the question of a quality, versatile Euro DDL to rest.


In most cases, reverb can glue sounds or make sounds more musical/pleasant to hear but for granular synth it can be must-have if end of loops/pieces are not jointless.

It can be a sad for a creator, that some modules are used for 5-10% of overall functionality/possibilities, but it’s the same in many domains. Like players who use only 7th or 8th string on guitars because they sound with solid bass and are enough for rhythm rattles.


Thru, but radical choices can be liberating.

If the actual Clouds didn’t have a reverb, maybe the users wouldn’t use it as a end-of-the-line module that much. Not saying that is bad or anything, but it could affect the creativity.

Regarding the reverb, I will be considering to get a 2hp verb.


two points to mention:

  1. it’s not bad if thousands of people will buy new clouds just to hide their skills in music composition/sound design (like I can do :smile:). Trying to edify people to think and explore is very hard task and maybe a kind of the task Sisyphus.

  2. (very IMO) I very some doubts of 2hp reverb. Very cheap and small factor to sound good (main processor +ADC+DAC). Some examples that I like: BAM, Axe-FX II reverbs. Just for Axe, reverbs have a configuration Hi-Fi/Lo-Fi just to save some resources. And these effects are based on powerful and expensive Tiger Sharc.
    Cheap reverbs can ruin your sound.


Not sure how you translate this to the “internal reverb - yes or no” argument. The reverb built into clouds is extremely constrained to the small amount of RAM and processing power left on the microcontroller. I’m repeatedly impressed how good it sounds under these constraints (I still don’t understand how Olivier pulled this off with so little RAM, my own reverb algorithms always require external RAM…) - but ultimately the 2hp verb has much more “raw” power available with its FV-1 DSP (that is - btw - also used in tiptops Z-DSP).
So judging from available DSP power alone the 2hp verb could eat clouds’ reverb for breakfast.

Ultimately what’s far more important than raw processing power is the actual reverb algorithm = how the available DSP is actually used. All the power in the world doesn’t help if you don’t know what you’re doing.

Bottom line: I’m not saying the 2hp reverb is good. What I’m trying to say is - just because it isn’t expensive, that doesn’t mean it has to be bad.

(Side note: If you assign a “price” to the clouds reverb, it would probably be super low given that it only uses a quarter of a knob and just a fraction of the power from its 11€ processor. For the 10-20€ it probably costs proportionally to the full module, the sound quality is incredibly good. Good enough to be used as an end of chain reverb by many people even though it has a high noise floor and only a single wet/dry/size combo control. Again, that goes to say a lot about the quality of Oliviers work.
On the other hand, devices like BAM spend most of their budget on the case, power supply, general I/O, etc. and only a tiny fraction of the cost goes towards the actual reverb processing.
What I’m trying to say is: Again, price doesn’t tell anything about the quality of the reverb.)


If you are looking for an end-of-chain reverb, then I don’t think that you’d want one with a lot of CV control necessarily. End-of-chain means that everything is going to go through it and do you really want CV modulation to effect everything?
I find that using a Pico DSP or Clouds to add a specific, non-traditional reverb to particular sounds and then sending out the mix to an external pedal like a BigSky or H9 works better, at least for my workflow.


Clouds’ reverb was originally developed on an FV-1 (it’s fairly obvious from the code :D) and consumed about 25% of the resources available on that chip.

If you need a Tiger Sharc to run an algorithmic reverb, you’re doing it very wrong. BAM is just a dsPIC, I’m not saying that to diss the product – instead, that you don’t need much to make things sound good. Crude interpolation and lower resolutions (12-bit) are known to give reverbs a more pleasant character, so hi-fi and technical sophistication might go against quality.

Regarding RAM… A room would have to be very reflective or very large if sound rays traversed more than 600m before reaching you. That’s only 1.75s worth of delay RAM. You can also cram 10x 60m-long rays in that, or for small rooms 30x 20m-long rays.


Is only a single delay buffer required, for a reverb effect?


It depends on the architecture. I’m not aware of any algorithm that uses just a single delay, but maybe you know one, Olivier?

To my knowledge, the two most used algorithms are

  • Lexicon-style reverbs, which need two main delay buffers for both sides + some more delay buffers for the schroeder allpass diffusors of which you can use multiple in various places as you wish. The two main buffers are connected in a figure-8 pattern.
  • Feedback delay networks, which are exactly what they sound like: A network of multiple delays that feed back into each other. The individual delays are usually mutually-prime in length. To me, it’s a little bit of a black magic art to make this sound good, but it can be made very good sounding.

You can spice up a basic algorithm by

  • Adding a model of early reflections (they will most likely be coming from a delay line with multiple taps),
  • Adding schroeder allpass diffusors in various places to increase the echo density (each diffusor needs another delay line)
  • adding granular readers to the delay lines to make the echo density even more unpredictable, but you’ll have to be very carefull to limit the total gain in the reverb tank, otherwise reverberation times will become unpredictable quickly.

Bottom line: There are many places where delay lines are needed in an algorithmic reverb. How many you’ll need in total depends on the “creative choices” you make when you design the reverb algorithm.

Some links I found interesting:


I’m not aware of such algorithms either, but my point is that once you convert delay RAM time to length, you realize that you can already do a lot with a moderate amount of delay RAM. The FV-1 only provides 1 second of delay (340m).


But can we expect some MI effects? :slight_smile: or is this something you are currently not interested in developing.


Who knows!


I totally agree with you that algorithms (in terms of their right coding/musical “taste” of developer) are much more important that used components. I just make a link how module sounds to its restrictions (of module size/costs).
… and have some doubts of putting to the end of chain not top sound (only for my ears) modules.