I’m not sure if this is a plugin specific thing, or a system-wide modification, but there are a fair few situations in which hitting a plugin causes a digital click. If there’s a delay or reverb after the click, it gets amplified and repeated (which I guess means we know it’s happening at that point in the signal path, rather than just as a computer glitch affecting the outputs…)
Is this a known thing that’s being worked on? It’s not crippling, and for the most part, it only happens if there’s audio going through the plugin at the time but it’s making a few of the more experimental things I’m trying to get the Duo to do difficult to use…
On a related note (positively) I’ve been doing a lot of very detailed listening/testing over the last few days, both performance wise and listening to recordings, and the basic fidelity level of the box is wonderful. I can’t wait to see the plugin set mature and grow - it’s extraordinary just as it is, but it’ll be unbeatable soon
I’ve been spending about 6 hours a day programming and tweaking sounds… exciting times!
I’ve heard clicks before, most notably all amps or anything that changes gain.
Since most plugins are simply hard-bypassed, a click is normal.
Click-free bypass needs to be supported by individual plugins on a plugin-by-plugin basis.
The x42 Parametric Equalizer does this for example: The MOD’s hard-bypass button is not available on the GUI.
The En/Disable control does not 'bypass" the plugin but instead internally smoothly fades all controls to “0” click-free.
In guitarix, the bypass button is as well provided by the host, we do it with host based ramping (dry/wet smoothing before de-connect), instead hard bypass.
Yep, x-fades or simply fade in/out is a common technique. It really depends on the DSP in question which is why this cannot be host-provided, but needs to be done by the plugin.
In some cases (eg. EQ or delays) dry/wet can lead to audible comb-filter artifacts which defeats the purpose of smooth bypass. There is an extension (port designation) underway so that a host can inform a plugin about bypass/enable.
I don’t think we should expect all existing plugins to implement new stuff which is not really required.
Since MOD runs at a fixed buffer size, perhaps a smooth bypass over 128 samples can do the trick for most plugins…
Some plugins will need to do the bypass themselves, but I think that’s a small number of them.
In my case, all x42 plugins that do either synthesis or DSP are affected when hard bypassed.
The meters, scope and tuna are fine since they just pass-thru. Midifilters could also make use of an explicit bypass (to flush queued events, e.g the midi-delayline or note-toggle) to prevent stuck notes. Cutting of synths while they’re playing a sound will also click.
If a plugin modifies the audio signal, and you hard switch, there will be a jump in 1 sample, depending on how far it jumps the resulting click may not be audible. Then again an immediate 1-sample transition in timbre can also be unpleasant.
128 samples at 48kHz sample rate corresponds to 375 Hz. That’s an audible fade and it will click for all signals less than that. Note that a guitar’s A-string is 220Hz and a Bass Guitar goes as low as 29Hz. Even if it’s not a hard 1 sample click,a rapid fade will introduce audible harmonics in that case.
For gain a time-constant of corresponding to 25Hz for a 90% exponential fade is common.
we use
steps_down = (256 * samplerate) / 48000;
for ramp down and
steps_up = 8 * steps_down;
for ramp up, those have proven to work well for 99% of all plugs, (leaving 1% for future use )
previous value for ramp down was (64 * samplerate) / 48000, but some plugs (reverb, delays, etc.) still issue with that.
From the users view, you wont notice the ramp time.
@brummer: That’s awesome. More professional plugins!
For guitar/bass FX, fhat seems perfectly sane. Assuming you’re near the “peak” (1.0) of a low bass note, steps_down would add an overtone at 187Hz while ramping. steps_up is pretty safe (~23Hz). But yes it depends on the DSP.
Those are for guitarix provided bypass controls, right?
If the MOD-host hard-bypasses a plugin, the plugin cannot know, which is what the new lv2:enabled property is for.
That are for the bypass controls provided by guitarix as host. The Gx plugs didn’t provide bypass ports. We assume that bypass should be handled by the host.
Those values have been proven over the time to be save for us, true, during ramp down, theoretical artefacts been possible with this, but in practise they don’t happen.
As a side note, the GxAmps provided you, in the first incarnation, the same popping noise you get when you switch on a analogue amp with gain turned on. To avoid that, we’ve added a new value to the tube emulations (vk0) which ramp it internal down. But this is just for the internal engine in the amp simulation and has nothing to do with the click you will get when hard bypass/enable.
Very glad this discussion is happening. The technicalities of how you fix it are fascinating, but not something I can usefully contribute to except to test them out as a performer - gimme a shout if you need to try some things out
FWIW, the digital pedals I have that offer true bypass or buffered, I almost always go with buffered so I can keep the tail on any time-based effects… But none of the true bypass pedals produce a click… Likewise my Lexicon MPX-G2 didn’t give any audible clicks when switching effects blocks on and off…
It may not seem much to you, but just reporting the issue and providing an example that reliably re-produces the issue is already a very valuable contribution! Thanks for getting the ball rolling on this one.
)