Inconsistent clicks (analog machines)

I am sorry if this is too off-topic (happy to delete, in case): the clicks really bother me on the Rytm and it seems to me that the Syntakt is almost click-free in comparison. Like the sounds’ attack components are a bit too loud or pronounced in comparison to the body on Rytm.

I had a similar thing happen with the tone synth. It would have random clicks in the beginning, no matter how much attack i would dial in. I don’t know if it’s caused by the setting of one of the synth engine parameters or the amp/filter settings. It’s a relatively quiet click but still there and not on every note. I’ll try out some more options to see if i can solve the problem.

This person is leaking the SynthTone already! #confirmed
/s

2 Likes

OK, so after some more testing I think I have found the issue!
The clicks happen when the decay parameter from the synth page is longer than the length of the amp envelope. It seems that when the amp envelope re-opens there is still a little signal left behind from the previous trigger. I was using the AMP envelope as the ONLY envelope to shape the entire sound, but it seems that I also need to consider the decay in the machine too. It is a little disappointing as the amp envelope shape will now be different as the 2 envelopes combined will create a different shape. It especially impacts the use of the hold stage. I will report this to Elektron as a bug, as NO sound from a previous envelope should be leaking into the next envelope. At least now I know what the issue is and I can learn to navigate around it. Not a massive deal, things still sound great :blush:

short version
If anyone else is getting clicks simply reduce the “DEC” parameter on the “SYN” page until the click has gone. Hope this helps.

22 Likes

thanks for the hint!

Nothing’s leaking anywhere, it’s definitely not a bug, it’s the nature of what you are trying to do without accepting the simple realities as I’ve stated above and I’m pretty sure Elektron will advise that there’s nothing they can do (as that necessitates compromises, at best)

Here’s why it’s generally only ever discussed in the context of low frequency or simple fundamental waves … it’s because it’s just so much more obvious

You are triggering an envelope whilst the oscillator is currently sounding a low frequency wave, that existing wave can be at any position in its phase cycle, so it’s immediately reset to zero, this means a potentially massive change in the existing periodicity at the transition and the typical evidence of this is shown below, with and without oscillator reset for both the ‘saw’ and ‘sin’ - this is a single oscillator from the DVCO, no filter and amp fastest/longest i.e. fully open. Only adopting to use SYN decay to show it doesn’t change the reason clicks occur, it just avoids the issue by taking away the key ingredient, which won’t always be an agreeable compromise, especially for the DVCO or long kicks etc

The issue I was discussing above relating to the envelope profile is no different whether it’s the amp envelope or the internal ‘fade’ settable per oscillator

It’s all about what happens when the phase is arbitrarily reset in an instant, there will almost certainly be a very obvious change in the frequency content due to the truncated wave which is implicitly short relative to the interrupted content … shorter waves are higher frequencies and as it’s always less than a cycle and probably between two lower frequency ‘neighbours’ it is conspicuous too - this is just not as noticeable with higher frequency ‘busy’ content, but it happens there too, you just notice it way less and as the content frequency is higher it can be that the ‘click’ interruption is of a similar frequency anyway

Further, slow AMP envelopes will NOT address this, as you are still going to abruptly cut the existing wave at an arbitrary phase point and therefore truncate a low frequency cycle into a higher frequency cycle

This is why there is a so called ‘analog’ envelope option which does not reset the amp envelope, so the envelope builds from where it is, this can result in less severe interruptions as the volume profile isn’t being as radically transformed and may be less conspicuous irrespective of any oscillator phase reset

I’m discussing this in detail because my points are missed and it helped when doing this for OT transitions when streaming from card or ram at loop points, it’s the same basic physics - you’re interrupting a wave and implicitly truncating it in some way, therefore creating high frequency blips

The reason it is ‘inconsistent’ is because the phase is completely arbitrary when the oscillator is triggered (if not resetting) or retriggered (if not or even if resetting)

This isn’t a bug and the workaround is only making a difference because you kill the source sound, so there’s no wave to ‘interrupt’ - it obviously only works if oscillator reset is enabled, but it’s no use for the DVCO if you want sustaining notes (you don’t want any SYN decay) and it’s no use for notes close together as you still have the same basic fundamentals crashing the party - interrupted low frequency waves are going to click

Here are a few pictures which always help clarify the thoughts

There’s a set for DVCO ‘SAW’ and ‘SIN’ with Phase reset on and off and also where notes follow one another, in this case octave jumps up

There’s a clarification diagram at the end showing how the phase of the wave you interrupt is completely arbitrary and therefore introduces unavoidable clicking, the theoretical mitigations for this are unpalatable and this is why it’s just something (especially on an analog circuit) we just need to accept as the nature of the beast

This is one of those topics that people come back to time and time again, insisting it’s a bug without considering what is actually happening, I hope this post goes a little way to understanding why there is inconsistent clicking and when it’s most noticeable


SAW FREE

SAW PHASE RESET

SIN FREE

SIN PHASE RESET


PHASE ‘JUMP’ POSITION IS ARBITRARY : Existing waves will generally click

Consider the truncated (dashed triangle) wave-period ahead of the reset phase oscillator with ANY amp profile - you generally create a ‘higher frequency’ perturbation that will be more or less obvious depending on when the new note is requested … the lower the frequency either side of the transition, the more obvious, the interruption impact is arbitrary (thus inconsistent) and largely unavoidable

ANY AMP profile to the right of the trigger point, whether reset or not, will not change the fact that you have ALREADY forced a premature termination of a wave which is implicitly shorter and will be conspicuous in low frequency simple content. This is most evident with content which is tending towards being of a low with simple fundamental only nature, therefore the contrast of this ‘diverted’ wave is simply more obvious than it would be in almost any other circumstance.

A bug is unexpected behaviour, this is not a bug

28 Likes

I understand your explanation, thanks for taking the time to explain things (although I must say you seem angry about it, with the bold and caps etc :rofl: I hope not, maybe I just read it wrong :blush:)

At the moment a new envelope is triggered the waveform is suddenly changed from one state to the new (reset) state causing a click. Inconsistent because there is no way to know at what point the wave will be when the envelope is triggered. I understand this. But what I do not understand is this:
When a new envelope is triggered the signal level is momentarily at silence for the length of the attack time until the level is brought up to full volume. Even if the attack time is set as short as possible it does not start at full volume, rather it starts at silence and then as fast as possible rises to full volume. If the amp is silent at the moment the wave is reset, how is the sudden change in wave shape even audible?

For example, if attack time is set to a value of 2 (not sure what this equals in millisecond) then the envelope is definitely starting with silence and then very quickly rising to full level. The time it takes the envelope to fade the level up is after the moment when the wave was cut, so how can it be audible?

Also, I still do not understand how exactly the envelope reset feature works. You referred to it as “so called analog”, where did you get that from? I do not see a mention of this in the user manual, is it referred to this way in a different manual perhaps? If the amp envelope is halfway through the decay stage and a new envelope is triggered (with the AENR turned off) will the new envelope simply ignore the new trigger and finish its decay stage as normal, or does the new envelope jump back to the start of the decay stage again? If you know how it exactly works, please explain it to me (if you don’t mind)

The last thing confusing me here is how is this related to analog only? Surly a digital OSC could also be at any point in its wave cycle when suddenly asked to reset to zero phase and this would cause the exact same clicking behaviour, right?

I tried all my other synths for the same behaviour and I am not able to replicate it.

1 Like

it appears you haven’t quite even read it - the bold/caps is an emphasis (amidst a dense text that is probably quite repetitive, to make the point register), had you read what was emphasised (rather than noted the emphasis) you may not have needed further clarification

it’s explained enough imho to be useful to those curious/interested and as a springboard for further thought … anything that needs extra reflection can surely be googled

it’s already frustrating to have to witness these things recirculate, be it on DN, OT etc and have it discounted as a bug

There’s plenty of food for thought posted here/elsewhere, it’s perhaps an idea to reconsider some assumptions as to why it’s maybe not a bug and just a factor in the nature of the content and that it may involve compromises

There may be mitigating aspects that can be incorporated, but as you can’t predict the future, nor necessarily delay what’s present to the future just to make its arrival/transition more seamless or transparent - it’s the nature of the problem that you can’t have it all, not on a monophonic voice, not without some compromise and mitigating for it is obviously much easier in the digital domain

It’s all relatively easy to try out and that (along with some visualisation perhaps) is the best way to examine it on your own terms - especially if using a sonograph or spectrograph for all the nuances being considered

4 Likes

6 Likes

Of course I read it :slightly_smiling_face:
Maybe consider making your text less “dense” and “repetitive”
I also did not ask for further clarification on the points you raised, rather asked questions that were not covered in your explanation.

It seems you are exiting the conversation now and are not willing to answer my other questions. OK, thanks for your time anyway :+1:

I have to disagree with you about this being the nature of monophonic synths as my other synths do not exhibit this behaviour. This is in fact the nature of Elekton synths.

Newly triggered envelopes should start at the OSC reset point (which is silent) and not include any audio from the previous part of the wave (before or during the reset). It seems to me as though the envelope is triggered slightly before the wave reset. If it were triggered after then no sound from the wave reset could be heard.

I will continue the conversation with Elektron support regarding if this can be improved.

2 Likes

The bold ‘Angry’ bit for context here

I’m repeating what needn’t have been repeated had the text been considered rather than dismissed because it doesn’t fit a prior assumption.

2 Likes

Yes, I read this. I agree that it doesn’t change the fact that the wave was reset prematurely. But that is not what I am debating.
I am stating that a new envelope and a newly reset wave both starting from silence should not have any audible artifacts from the wave shape being prematurely interrupted.
If the new amp profile was truly starting at exactly the wave reset position then there would not be a click.

What you’re hearing is not the new envelope and the newly reset wave. The click comes from the previous wave that is abruptly being cut off.
If a wave is not at zero amplitude, it doesn’t matter if you start or stop it in an instant, it will click anyways.

12 Likes

Oh, I see that now, thanks.
I was thinking that the click sound was part of the start of the new envelope rather than the end of the previous one 🤦
I’ll have a think about this :thinking: thanks.

1 Like

Glad it’s clicked now :wink:

14 Likes

So to avoid clicks, you would need to carefully set the decay of your envelopes.

If we one day have the opportunity to setup a custom default Kit, I will definitely make all amp envelopes with finite decays!

3 Likes

I do still wonder how it is achieved in other synth designs. I tested with my Novation Summit (it has OSC reset) in mono mode and was not able make it click. They must be designed differently in some way, no idea how. But I do think if it would be possible to improve it somehow for Elektron?

And I still do not know why only the analog machines do it. A digital OSC that is abruptly reset should have the same artifact right?

An answer to this would explain why I can not get my Summit to click, as that has digital FPGA numerically controlled OSCs.

Sadly my PolyBrute does not have wave reset so I can not test there. But I will probably test my Tempest tomorrow and see what happens there. Oh and Pro 3 doesn’t click with OSC reset either.

1 Like

The nitty gritty is ultimately in how the descent from the truncated wave is handled. If it’s slewed on its path back to zero it will change the nature of that impulse and reduce the frequency content. But in doing that you necessarily introduce a sort of latency to the incoming new note’s onset. Maybe this is no big deal with a synth sound, but in the context of a drum machine it may be better to prioritise timing. So as you can’t consistently predict what’s coming then the reality is that you will get a very fast transient at that transition just ahead of the new note.

If you want to understand how synth X and Y handle it differently, throw two similar sounds into audacity or similar

So the energy of the transient comes from the existing wave being shut down too rapidly, arguably too well for some. It’s okay to micromanage this to taste by muting the source sound ahead of retriggering.

Some other gear may mask this, but as I said originally, this mitigation also has its own cost. And as you see in the analog oscillators there can be a bit of DC offset and a reset wave is not pristine like a modelled numerical representation, so mitigating this properly may simply be trickier.

It’s still not helpful to think of it as a ‘bug’, it’s sounding like it should, given what’s actually happening. I’ve not been explaining this exhaustively either, (i’m not able to) but there should still be some useful things to glean from some of it to dispel any ideas that it’s something that isn’t already understood.

A more flexible ‘compromise’ is to turn off both oscillator reset and amp envelope reset, this approach where the oscillator phase is not reset benefits from a smidge of attack to ease it in, but it’s less clicky, but also less predictable, especially if leaning on fm or ringmod etc - but if the ambition is clickless low freq sounds, then it may be your best option as the note isn’t in a rush to define itself

FWIW I have no issue making almost all digital source oscillators click, but strangely they do respond differently (i.e. are mask-able) to their low pass filter, all of which makes me wonder if it’s those analog vagaries again at play somehow or a function of the voice topology or filter design

So with careful use of what’s at hand, you can get quiet dubby subs and clicky kicks to suit broad tastes, but your success will be tempered by details like the wave/env/filter profile both pre/post for any given transition

If you need osc reset, then that may require some compromise. Or embrace the click

I’m more curious about understanding the differing responses of the filters, if they’re even closely conceived

2 Likes

Interesting, thanks.
I just realised something. When I was not able to get Summit to click I think I realise now why. One of its mono modes cycles through the voices, so that explains why it wasn’t clicking haha, oops.

2 Likes

It’s easy to forget details when you have too much gear and everything has its own idiosyncrasies (or if you’re just gettin’ old like me) - but the old abrupt reconfiguration of a low frequency wave making clicks is more normal than not in my experience

Elektron made strides to appease DN owners bemoaning clicks, so you never know ! But my gut feeling is that this is something to work with/around

3 Likes