Interpolate controls to eliminate zipper noise about dpl.lv2

[unfa]

This is exactly what I've done:
https://user-images.githubusercontent.com/491704/125860451-8d8262ac-e4ec-4e67-8d06-145ef7ca38de.mp4

[x42]

yes, it was too fast. Now it's an exponential approach that reaches the 90% of the target value in ~10ms.

Still I'm not too worried if a rapid +/- 40dB control value increase introduces some audible artifacts. Going from min to max in a short time is borderline insane.. still thanks for bringing this to my attention.

sure, please email the video to me.

[x42]

The input-gain is filtered, however the threshold is not, default release-time applies to it though.

That is intentional: no sample must exceed the threshold, ever. If it was low-pass filtered there may be short times where the signal can exceed a given threshold.

I assume you've automated the threshold. What use-case is there for that?

[x42]

I see. The issue is sparse automation. Internally the de-zipper takes place over a very short time (sample-rate dependent; at 48kHz the fade take 64 samples), but Ardour only sends a new value every cycle, here 512 samples / 48kHz = 10ms ~ 1 horizontal gird in the scope:

So there's a short ramp (internal dezipper) and then a longer flat period.

[unfa]

I actually just changed it quickly by hand while playing loud bass though and noticed artifacting.
I would very rarely automate a limiter, but at times such option is useful if I for example want to remove dynamics on a drum part for section of a composition to then bring that back.

[x42]

Which parameter have you changed?

[x42]

git pull :)

Sadly just missed the quarterly release cycle by less than 24hours.

[x42]

now ramping up 30dB over the span of 1/2 second at 1024 / 48kHz looks like

One process cycle (new gain value) happens roughly every two horizontal grid units.

[unfa]

Alright, so the interpolation was just too fast? :D

I'll remove the mention of this in my video then to not confuse viewers.

BTW, would you like to watch my whole video before I publish it to make sure I'm not getting something completely wrong?
it should be about ~10m long total.

[x42]

Fixed in dpl-1 v0.5.3 -- which is part of the x42-plugins 20211012 release