Presswerk (most compressors) producing low frequency content

[OdoSendaidokai]

Hi,

this is more like a general question because i noticed when doing a low-cut even rigorous (8 pole at 700Hz) before Presswerk, I get lots of low frequency content after using Presswerk. This looks like a normal behavior for most compressors, except the spectral compressor from robbert (yabridge).

Now I'm interested on the "why does this happen" and does this mean i have to low-cut again every time after the compressor? ... because of Fletcher-Munson and the energy distribution...

I have difficulties to find the correct words for doing my own research about this topic. Could someone point me to some explanations, papers or the correct word to search for?

Thanks in advance.

[DMG68]

Google ‘sidechain filter compressor’ to understand this better.
However, with Presswerk you can turn on the sidechain monitor to hear the filtering as well. Don’t remember if this bypasses compression though, probably not, some simple knob turning should tell you. Maybe check the manual as well, it’ll help explain it.

[Viktor [TUC]]

Hi Odo,
I'm not noticing any unexpected low end frequencies when using Presswerk. Can you say if perhaps that happens with specific settings only?
Also don't forget that you are of course compressing, so depending on your settings, low-volume bits are being pushed up so that they become audible, or at least visible in a spectrum analyzer. Low frequency content, previously not visible, becomes visible, if there's a low dB threshold in the spectrum analyzer.

Viktor

[OdoSendaidokai]

Google ‘sidechain filter compressor’ to understand this better.
However, with Presswerk you can turn on the sidechain monitor to hear the filtering as well. Don’t remember if this bypasses compression though, probably not, some simple knob turning should tell you. Maybe check the manual as well, it’ll help explain it.

Thanks but Sidechain is a whole completely different topic. It is not about sidechain that is left out completely. It is just about standard compression, with no extras.

[OdoSendaidokai]

Hi Odo,
I'm not noticing any unexpected low end frequencies when using Presswerk. Can you say if perhaps that happens with specific settings only?
Also don't forget that you are of course compressing, so depending on your settings, low-volume bits are being pushed up so that they become audible, or at least visible in a spectrum analyzer. Low frequency content, previously not visible, becomes visible, if there's a low dB threshold in the spectrum analyzer.

Viktor

Yes I already thought as well on that, but shouldn't a filter or EQ, especially with a very steep slope avoid that? Here is a screenshot with two different EQs after the compression. The display in EQ+ is a little flattery and it is difficult to screenshot it, but there is as well such a nose to the left.

One Idea I received was, that this could maybe be subharmonics. I found some things about it, but it is not clear to my, why a compressor should produce them.

Edit: A second Idea are intermodulations ... will have to do my research, to find a good explanation why this could be the reason.

01_temp.jpg

[DMG68]

Google ‘sidechain filter compressor’ to understand this better.
However, with Presswerk you can turn on the sidechain monitor to hear the filtering as well. Don’t remember if this bypasses compression though, probably not, some simple knob turning should tell you. Maybe check the manual as well, it’ll help explain it.

Thanks but Sidechain is a whole completely different topic. It is not about sidechain that is left out completely. It is just about standard compression, with no extras.

I misunderstood, I thought you were using Presswerk’s filter. I’ve never experienced what you’re mentioning either. Can’t help you out, you’re in good hands with Tuc.

[Viktor [TUC]]

...shouldn't a filter or EQ, especially with a very steep slope avoid that?

Looking at the screenshot, the EQ slope is a 12dB/oct slope, so not particularly steep. I'd absolutely expect low frequency content to make it through.
The spectrum display in that EQ+ instance also shows low frequency content in the original signal, so it doesn't look like being produced by Presswerk.

Your bigger spectrum display (TDR Prism it says), what does it show when you bypass Presswerk in that scenario?

Viktor

[gaggle of hermits]

OP, what are you actually trying to achieve here?

if you attempt to filter out a frequency range before a compressor, you are pushing the focus of the compressor onto the remaining signal, so it will push down harder where you have the higher frequency content in this case. once you add makeup gain, you wind up pushing the low-frequency material higher again unless you've got a brickwall filter ahead of the signal – which you haven't.

this is why dmg68 wasn't off-target talking about the effect of the sidechain signal, because that's one of the reasons for putting an EQ on the sidechain - it focuses the compression on a particular part of the frequency range. in your case, the compressor still has a sidechain signal going in here, it just happens to be the same as the source.

don't want those low frequencies? put the filter *after* the compressor.

[Viktor [TUC]]

Edit: A second Idea are intermodulations ... will have to do my research, to find a good explanation why this could be the reason.

Good idea, but I don't think it applies here. Intermodulation distortion occurs when you send several signals through a non-linear device, like something with distortion or saturation, which yes Presswerk does offer, but your screenshot shows that the saturation section at the bottom is inactive.
Try this: Send two sine waves into Presswerk, turn up the 'Amount' knob at the bottom in the Saturation section, look at the result in a spectrum analyzer. Begin with the sines at same frequency, then start moving them apart in frequency, you'll see how high and low "side bands" are generated after the distortion part. So you can get frequencies lower than your original signal.
But like I mentioned, your screenshot suggests that there's no distortion happening here.

Viktor
u-he team

[OdoSendaidokai]

Edit: A second Idea are intermodulations ... will have to do my research, to find a good explanation why this could be the reason.

Good idea, but I don't think it applies here. Intermodulation distortion occurs when you send several signals through a non-linear device, like something with distortion or saturation, which yes Presswerk does offer, but your screenshot shows that the saturation section at the bottom is inactive.
Try this: Send two sine waves into Presswerk, turn up the 'Amount' knob at the bottom in the Saturation section, look at the result in a spectrum analyzer. Begin with the sines at same frequency, then start moving them apart in frequency, you'll see how high and low "side bands" are generated after the distortion part. So you can get frequencies lower than your original signal.
But like I mentioned, your screenshot suggests that there's no distortion happening here.

Viktor
u-he team

At the end there is distortion and/or kind of saturation going on, because I'm using threshold, ratio, attack and release. So Intermodulation could be possible.

The whole topic is not a Presswerk specific topic. It is surely a topic for all compressors of this type and I'm also sure, it is not a bug. I just want understand WHY this happens and IF it has a negative influence on my sound, I wasn't aware of before.
So i hope I find here some people, who know why this happens, what's for example on the screenshot.
AND .. do I really need an additional EQ after the compressor, if I want to preserve a clean mix?

[Urs]

Transients appear to be low frequency content on FFT-based (and possibly all) spectrum visualisation tools. A compressor changes transients and might therefore appear to have an impact on the visualisation of low frequency content.

My guess.

[Viktor [TUC]]

I just want understand WHY this happens ...

I'm not trying to pick a fight so apologies if I come across as pedantic, but we have yet to establish THAT it happens, your screenshots at least do not show that. They don't show what the frequency content of the original signal is, for starters. They do show a gentler EQ curve, suggesting that whatever was there before will not be fully removed.
When I test it on my machine, I don't get any surprising and unexpected low-frequency content after Presswerk (leaving the Saturation deactivated, which can add intermodulation distortion depending on what you send in).
With a highpass filter of a 12dB/oct slope (like in your screenshot), I could set it high up to 10kHz and still have things happening down at 40Hz, that's why I'm not surprised it's there even after compression. It's just inaudible.

My example image: Highpass-filtered drumloop, filter set to roughly 1kHz, very steep filter (48dB/oct).
Visible frequency content down to 400Hz, and the graph has a low limit of -130dB, so that's way beyond audible.
(If I lower the limit to -180dB I even get rare bumps down at 40Hz in the analyzer.)
Left side, no Presswerk. Right side, Presswerk doing lots of compressing. Nothing is added.

HP-filtered Compression.jpg

[OdoSendaidokai]

I just want understand WHY this happens ...

I'm not trying to pick a fight so apologies if I come across as pedantic, but we have yet to establish THAT it happens, your screenshots at least do not show that.

No no, I didn't took it as a fight. I'm only try to understand and learn something
And maybe I didn't get the connection from my point to your explanation.

They don't show what the frequency content of the original signal is, for starters. They do show a gentler EQ curve, suggesting that whatever was there before will not be fully removed.
When I test it on my machine, I don't get any surprising and unexpected low-frequency content after Presswerk (leaving the Saturation deactivated, which can add intermodulation distortion depending on what you send in).
With a highpass filter of a 12dB/oct slope (like in your screenshot), I could set it high up to 10kHz and still have things happening down at 40Hz, that's why I'm not surprised it's there even after compression. It's just inaudible.

Hmm ok, the bitwig EQ shows the input and output of the signal. So the shadowed signal is the input and the white line is the output of the EQ.

01_temp.jpg

I uploaded the video we discussed in my discord community .. maybe it shows a little better with what I'm struggeling. It's german, but I think it shows my "problem".

That it is inaudible is ok and I wouldn't be "concerned". It's just the energy consumption of frequencies that are not needed and I'm curious why this happens.

https://www.youtube.com/watch?v=-URAfo9MwDc

[Viktor [TUC]]

Hi Odo,

the misunderstanding is that the analyzers you refer to all stop at -30 dB. They don't show quieter signals at -50 or-80 or-100dB. In the video you mention that there is apparently no low frequency content, but that's just not the case. There is, but those analyzers don't show it because it's below their -30dB threshold.

Use a big analyzer that shows you frequency content down to -120 dB and you'll see that even after your highpass filtering, you still have something going on at those low frequencies, it'll just be very quiet.

Viktor

[OdoSendaidokai]

Hi Odo,

the misunderstanding is that the analyzers you refer to all stop at -30 dB.

No that's wrong from what I see in that video.

I show with Prism and Span and they going very deep (-120db) . Or you mean a different deep.
And the Bitwig EQs, even they aren't going that deep (only -30db) are showing the low frequency content too.

I'm not sure on what you are looking? Or if we talk about the same video?