Latest News: u-he releases Sugar and Spice for Hive 2
Presswerk (most compressors) producing low frequency content
-
- KVRian
- 702 posts since 18 May, 2007 from Berlin
Hi Odo,
ah true! I stopped at where you pointed at the small analyzer windows and said that there's no low frequency content anymore. I see what you mean now in the big analyzers.
You have the filtered signal sloping down until they're out of range of the analyzer - you could extend the analyzer range to see more, Span can go down to -180dB, but then again a 24bit signal has a dynamic range of 144 dB, so whenever you see "energy" at that level, that's possibly misleading because it's just some noise that a circuit introduces perhaps, or the aforementioned intermodulation distortion (let's call it IMD from now on, too long to type), which (with complex input signals) may as well be just some noise across the spectrum that's getting added.
Be aware that noise at such low levels doesn't contain any relevant energy. Analyzers can make it feel rather big, but imagine the smallest possible value that's non-zero in a 24bit signal, occasionally blipping in and out (i.e. noise), but the analyzer paints a huge green block that suggests big deep rumbling sound.
You know, the fact that I didn't get that low level content on my first tries tells me that this is most likely IMD, because my test signals were rather clean and had no uneven harmonic relationships, and therefore didn't produce that kind of noise. That supersaw chord sound that you're using in the video, that's a proper complex signal.
Testing again with a comparable sound, I get those noisy results too. And if I add a lowpass filter to the highpass filter and get a super narrow band of signal, I can see that noise is being produced on both sides, low and high frequencies - to me that's the resulting noise from IMD, looks like close to pink noise. Depending on compressor settings, that noise can be lower or higher, but noise at the -100 dB level or even lower shouldn't be a concern when it comes to eating up energy.
In summary, I see what you mean! We both used different test signals, resulting in no such noise in my case, and some noise in your case. You get some pink-ish noise due to IMD with a complex signal, interesting, I learned something here.
Anyway, that's probably all the answers I'll ever have on the topic, I hope your curiosity is somewhat satisfied, Odo, I certainly enjoyed this!
Cheers and good night,
Viktor
u-he team
ah true! I stopped at where you pointed at the small analyzer windows and said that there's no low frequency content anymore. I see what you mean now in the big analyzers.
You have the filtered signal sloping down until they're out of range of the analyzer - you could extend the analyzer range to see more, Span can go down to -180dB, but then again a 24bit signal has a dynamic range of 144 dB, so whenever you see "energy" at that level, that's possibly misleading because it's just some noise that a circuit introduces perhaps, or the aforementioned intermodulation distortion (let's call it IMD from now on, too long to type), which (with complex input signals) may as well be just some noise across the spectrum that's getting added.
Be aware that noise at such low levels doesn't contain any relevant energy. Analyzers can make it feel rather big, but imagine the smallest possible value that's non-zero in a 24bit signal, occasionally blipping in and out (i.e. noise), but the analyzer paints a huge green block that suggests big deep rumbling sound.
You know, the fact that I didn't get that low level content on my first tries tells me that this is most likely IMD, because my test signals were rather clean and had no uneven harmonic relationships, and therefore didn't produce that kind of noise. That supersaw chord sound that you're using in the video, that's a proper complex signal.
Testing again with a comparable sound, I get those noisy results too. And if I add a lowpass filter to the highpass filter and get a super narrow band of signal, I can see that noise is being produced on both sides, low and high frequencies - to me that's the resulting noise from IMD, looks like close to pink noise. Depending on compressor settings, that noise can be lower or higher, but noise at the -100 dB level or even lower shouldn't be a concern when it comes to eating up energy.
In summary, I see what you mean! We both used different test signals, resulting in no such noise in my case, and some noise in your case. You get some pink-ish noise due to IMD with a complex signal, interesting, I learned something here.
Anyway, that's probably all the answers I'll ever have on the topic, I hope your curiosity is somewhat satisfied, Odo, I certainly enjoyed this!
Cheers and good night,
Viktor
u-he team
-
- KVRAF
- 6528 posts since 17 Dec, 2009
as far as i know nearly all non-linear audio processing, but especially those that saturate, produce low frequency content.
Nearly every saturator and compressor does that, and so do limiters.
how much and "is it a problem" really depends on the processing.
i can hear it well with my woofers so i doubt it's only visual
Nearly every saturator and compressor does that, and so do limiters.
how much and "is it a problem" really depends on the processing.
i can hear it well with my woofers so i doubt it's only visual
-
OdoSendaidokai OdoSendaidokai https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=466569
- KVRist
- Topic Starter
- 173 posts since 24 May, 2020 from Berlin
Yes the signals were different, because I wanted a more realistic scenario that happens more often. I will dig deeper here and try to find the reasons and as i mentioned in one of the first posts the intermodulation is a hot candidate for that. And I think dependent on the signal and another situation that lead to a problem. If i find it, I will discuss it in a video.Viktor [TUC] wrote: βThu Apr 25, 2024 7:16 pm
In summary, I see what you mean! We both used different test signals, resulting in no such noise in my case, and some noise in your case. You get some pink-ish noise due to IMD with a complex signal, interesting, I learned something here.
Anyway, that's probably all the answers I'll ever have on the topic, I hope your curiosity is somewhat satisfied, Odo, I certainly enjoyed this!
Cheers and good night,
Viktor
u-he team
So in my future workflow there will be the test with an analyzer and maybe another EQ after each Compressor I set, so I can ensure that Signal Is that clean how i want it to be.
However, I feel the same like you that I learned something very interesting and hopefully not that big interfering
-
OdoSendaidokai OdoSendaidokai https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=466569
- KVRist
- Topic Starter
- 173 posts since 24 May, 2020 from Berlin
Oh, ok so it would be good in most cases to check the signal after shaping to the desired range. Thanks!
-
- KVRAF
- 2346 posts since 20 Oct, 2014
I think the added low frequency content comes from the distortion the compressors add, if you use attack at zero and release at zero. At least both Bitwig compressors, "dynamics" and "compressor", will add distortion then. Pretty sure Presswerk does that, too, at 0 / 0.