NOVA and Q: dynamics detection vs EQ

[masitito]

Hi!

Yesterday I was trying for the first time NOVA because I was playing aroudn to see if dynamic EQ would help me on a track. Actually it did, but one thing struck me odd. The Q of the filter used for the dynamics detction is always lower than the one I can set for the EQ. I guess that this isn't an issue normally, but in my case it was too broad. By lowering the threshold the gain reduction occured too early.

Did you ever consider to decouple them completely? Or perhaps you can let the user change the ratio (offset?) between them.

Masi

[masitito]

In a video I saw a plugin by Fabfilter which seems to had an additional optional bandpass filter for the dynamics detector. As I understandt it comes after the standard filter for the band.

Masi

[FabienTDR]

This is really just a visual issue, both the SC and the bands are exactly the same. Their content/bandwidth is equal. We just draw them in different ways.

We're in fact considering writing a blog post explaining the point in detail. This is really a good question. Maybe soon on our website.

[vladg]

Hi!

Yesterday I was trying for the first time NOVA because I was playing aroudn to see if dynamic EQ would help me on a track. Actually it did, but one thing struck me odd. The Q of the filter used for the dynamics detction is always lower than the one I can set for the EQ. I guess that this isn't an issue normally, but in my case it was too broad. By lowering the threshold the gain reduction occured too early.

Did you ever consider to decouple them completely? Or perhaps you can let the user change the ratio (offset?) between them.

Masi

Hi!

It's really interesting question indeed. As the developer I can say that bandpass filters for both EQ and sidechain signal are the same. And on the first glance the curves look different because they have different scales (for EQ gain and for threshold values).

But the problem is that Q values for bandpass filter and for peaking curve are calculated differently (using -3 dB point for bandpass and 1/2 of peak gain point for peaking filter). It means in order to maintain the constant Q condition for resulting peaking curve (which is sum of bandpass curves) we adjust Q of bandpass filters so it's actually depend on EQ gain. But it doesn't depend on threshold! Which means Threshold curve on the display doesn't conform to "constant Q" condition. So that's why threshold curve has wider Q value (in sense of peaking filters), while Q value for bandpass filters is the same for EQ and for sidechain signal.

First, we're going to write blog post about it because it's interesting to know about Q aspects of bandpass filters in parallel equalizers.

Next, we're going to add alternative Q mode to Nova, which intended to maintain "constant Q" condition for threshold curve too. I think this mode would lead to more precise usage of dynamic bands. If this mode will sound better, we'll make it default.

We're preparing update for Nova so it's good feature to include.

[masitito]

Fabien and Vladm thank you for your answers.

I'm looking forward to your blog post. So far I'm overwhelmed by the audio engineering lingo. Currently I only fiddle knobs without knowledge about the math and physics behind it.

So probably I will benefit most if the visual scaling for the threshold can be toggled on/off in the UI. IMHO it is more confusing then helping, but this is certainly a matter of personal taste.

Anyway, NOVA is a great plugin. Thanks a lot for proving it for free!

Masi