|
|
|
06-24-2017, 12:21 AM
|
#1
|
Human being with feelings
Join Date: Apr 2010
Location: Cloud 37
Posts: 1,071
|
Highpassing Below the fundamental
So there's something I'd like to settle once and for all.
Some people believe that there is no useful information below the fundamental of a note.
Some people believe there is useful information below the fundamental of a note.
I'm talking about common Western instruments; clarinet, oboe, piano, guitar, violin etc.
In theory, it should make sense to use the steepest filter possible, and cut right below the fundamental, even automating to follow the instrument's pitch.
I know Surf EQ can do this automatically, but their High Pass isn't very steep. Neutron has a very steep High Pass, and I don't hear any resonance on it.
Reading a number of threads on the subject, it seems when people are saying not to high-pass, they're talking about weakening the fundamental (if I'm playing 'A 440 Hz' then yes, high-passing at 441 Hz can affect the tone, but cutting everything below 439 Hz should in theory do nothing but remove room noise, no?)
Reading a out sub-harmonics,
https://en.wikipedia.org/wiki/Undertone_series
It seems you can get undertones by over-blowing a wind instrument? So in the case I'm mixing an overblown wind, maybe I should be careful with the high pass?
Quote from Gearslutz
Quote:
ok im going to try to explain this the on way i can think of so please bear with me , lets say you record a guitar part the lowest string is bottom e 440hz, however when you play it back you can hear rumble and need to hi pass ,, is the low rumble considered the fundamental ? no , what has happened is you have recorded 3 or more acoustic systems the guitar body the guitar strings and the room all of witch have there own set of overtones and harmonics, so the guitar body could excite the room , these frequncies may be lower than the 440hz of the guitar but when you think about it the science is sound ,all thats happening is your hearing more than one acoustic source. the 440hz of your recording is still considered the fundamental
|
So there might be sub-harmonic room resonances. Are these musically useful if we want to keep the character of a certain room? I'm assuming if we're recording dry and creating our own room using reverbs, these are useless and can be cut.
I'm afraid of being misunderstood and going off-topic, so here's another example.
The low E of a standard tuned guitar 82.41Hz. I'm not missing anything 'musical' if I high-pass at 82Hz (in practice I'll just do 80hz to be safe). I know people will say it's an artistic choice etc... but if I wanted to include the guitarist's foot tapping I wouldn't even ask the question. In general there's no good reason *not* to high pass this at 82Hz.
Last edited by Mr. PC; 06-24-2017 at 01:36 AM.
|
|
|
06-24-2017, 02:18 AM
|
#2
|
Human being with feelings
Join Date: Mar 2016
Posts: 444
|
There are a few considerations here.
First, filters are gradual. So when you high-pass at 80Hz, you will be affecting 82Hz, however steep a filter you choose. The only question is how much you affect the 82Hz.
Second, filters are either minimum-phase, linear-phase or anywhere in between.
The minimum-phase extreme shifts phase so your useful signal will not sound exactly the same as before the filtering. Effectively, for a high-pass filter, lower frequencies get delayed relatively to higher frequencies. The steeper the curve, the more delay is added. Depending on the source, this may be actually a desired outcome or not. However generally this may make it harder to mix a live recording (where you have significant spill) because signals from different microphones are no longer coherent across the tracks. If they were not coherent to begin with (because of microphone acoustics or electronics, or differing HPF/EQ on the input) then this may not matter.
The linear-phase extreme does not shift phase however it comes with another caveat which is pre-ringing. Effectively the more a certain frequency is affected by the filter, the more it pre-rings. Pre-ringing is percieved as "muffling" of the sound because it affects transients. In the lower frequencies it may get to the point where it is very noticeable.
Both the linear- and the minimum-phase filters introduce post-ringing which is usually not very noticeable. Minimum-phase filters produce more post-ringing than linear-phase filters.
Third, whether or not you want to use a high-pass filter depends on the source material. If there is a strong room rumble or other low-frequency noise then you don't have any choice. If there are no such issues then there is not much point.
Fourth, it also depends on the source material in the sense that for some sources, low-end noise may be beneficial in some cases. For example, for some vocals, the chest noise may contribute to the perception of intimacy or emotion. For instruments with a percussive element, this may be also useful in some cases, for example a slap bass.
Fifth, from a mixing standpoint, you want to minimize the contribution of "uninteresting" frequencies in order to make space for the "interesting" frequencies. So when you have, for instance, a tom drum that is contributing low-end rumble but you also have a bass guitar and a kick drum then perhaps the tom drum rumble takes the energy that you could use in a more productive way. Or not, depending on the song.
|
|
|
06-24-2017, 07:26 AM
|
#3
|
Human being with feelings
Join Date: Apr 2010
Location: Cloud 37
Posts: 1,071
|
Ok, thanks for the reply.
In my situations I have 0 leakage, and record mostly close-micing in a dry room.
When I talk about high-passing at 80Hz, I really mean that I high-pass at a level that's low enough not to 'touch' the fundamental at all. For example I just set a low E from my synth and in order to stay below the fundamental I set the high-pass to 62Hz. I look and listen to make sure no difference is made to the tone.
About phase smearing - isn't that an argument against using any kind of EQ? Also, as far as I understand (maybe I'm wrong) if the high-pass doesn't actually effect the signal, it won't smear those transients. Is that right?
I would smear the sub-harmonic stuff (which I want removed) and possibly the fundamental if I put the high-pass too close, but really nothing else?
Is there a 'best EQ' in this regard, that alters the signal as little as possible? Somehow to my ears, Neutron just sounds way better than any other EQ I've used; I don't hear any negative change when I use its highpass filter (maybe a placebo).
Now I'm reading this,
http://ethanwiner.com/EQPhase.html
I'm thinking about phase issues from room reflections in a single mic - 1 more reason in my book to record as dry as possible.
Quote:
Third, whether or not you want to use a high-pass filter depends on the source material. If there is a strong room rumble or other low-frequency noise then you don't have any choice. If there are no such issues then there is not much point.
Fourth, it also depends on the source material in the sense that for some sources, low-end noise may be beneficial in some cases. For example, for some vocals, the chest noise may contribute to the perception of intimacy or emotion. For instruments with a percussive element, this may be also useful in some cases, for example a slap bass.
|
In most of my sources there is just some small room hiss, but since, as far as I can tell, there's nothing to loose by high-passing (I don't want to keep any 'noises' in these recordings) I might as well cut it out.
I just got frightened when I read it could cost my songs their mojo :| Seriously. So I just want to be clear that there's no mojo below the fundamental (and also that transient smearing shouldn't affect anything negatively... I know my link already states this; so I guess I'm just very not-confident).
Oh, here's something very cool,
http://www.falstad.com/fourier/
you can alter the phase of harmonics and hear how the sound changes... so maybe this really can harm the sound. Maybe we should avoid all EQ unless necessary.
Last edited by Mr. PC; 06-24-2017 at 07:32 AM.
|
|
|
06-24-2017, 07:47 AM
|
#4
|
Human being with feelings
Join Date: Mar 2016
Posts: 444
|
AFAIK the theory states that the less a certain frequency is affected, the less its phase is affected, too. However I don't have enough knowledge to make a certain claim so I'll fall back to personal experience. My experience is that linear-phase filters do something unpleasant to the signal. It may be right or it may be wrong. Phase smearing caused by minimum-phase filters is usually barely noticeable if at all. Still it is good to know it exists and may potentially cause problems. For example, if you do parallel processing on a certain track and then mix the outcome, you will get a comb filter effect when trying to use minimum-phase filters. Other situations include eq on reverbs or delays, mixing doubled tracks etc.
For a nice visualization of how minimum-phase filters affect the phase, you can use DMG audio's Equilibrium.
|
|
|
06-25-2017, 02:22 AM
|
#5
|
Human being with feelings
Join Date: Apr 2010
Location: Cloud 37
Posts: 1,071
|
Thanks.
I read that putting a minimum phase EQ on a reverb is fine because the reverb alters the signal anyhow, and so they're 2 different signals / no comb filtering.
Basically, comb filtering is the only problem that can be cause my phase issues, yes?
With a delay, if the delay is long enough, it also shouldn't be a problem, right? And if a delay is so short that it's getting into phase-problem territory, it could cause phase problems even without the EQ, right?
|
|
|
06-25-2017, 02:47 AM
|
#6
|
Human being with feelings
Join Date: Nov 2010
Location: Mullet
Posts: 829
|
Quote:
Originally Posted by avocadomix
For example, if you do parallel processing on a certain track and then mix the outcome, you will get a comb filter effect when trying to use minimum-phase filters. Other situations include eq on reverbs or delays, mixing doubled tracks etc.
|
does the comb filtering come from the further processing of the EQ'd track ?
i ask because using minimum phase filters in parallel does not cause comb filtering, so it must be from something else.
__________________
I like turtles
|
|
|
06-25-2017, 03:57 AM
|
#7
|
Human being with feelings
Join Date: Mar 2009
Location: Dartmouth, Nova Scotia
Posts: 11,251
|
Quote:
Originally Posted by Mr. PC
Maybe we should avoid all EQ unless necessary.
|
THIS is the most important thing you've said.
For instance, when you look at all the theoretical info, you get the impression you should filter just below the fundamental.
As you and others have stated, that doesn't necessarily work well with minimum phase EQ'a because of the inherent phase shift (aka group delay, time smearing, etc.).
BUT if you have. say, a loudspeaker crossover, a minimum phase EQ might be just the ticket to TIGHTEN things up as opposed to smearing.
WTF ???
Here's why:
A minimum phase EQ has the characteristic that as you adjust frequency response you also adjust phase response such that if the EQ is flat there is close to zero phase shift.
So when you EQ close to the musical fundamental you are adjusting BOTH amplitude AND phase.
This can work out OK, but often the results are not what was expected.
Now back to the crossover.
A single loudspeaker (say a woofer) is a minimum phase device.
It does not have a flat response.
Since it is a minimum phase device and it has a non flat frequency response, it MUST have a non flat phase response too.
Therefore by applying a compensating crossover EQ which is also minimum phase, when you correct the loudspeaker frequency response you ALSO correct its phase response, which actually tightens things up.
So a minimum phase EQ can be your worst enemy or best friend or somewhere in between spending on the circumstances.
Same goes for other filters, linear phase, excessive phase, all pass, etc.
So here's what I go by:
I solo the track and listen to it all the way through for obnoxious anomalies (foot stomping, etc.) and apply any filters I am forced to use JUDICIOUSLY.
Don't use EQ just because you own one, is the rule to live by
__________________
To install you need the CSI Software and Support Files
For installation instructions and documentation see the Wiki
Donate -- via PayPal to waddingtongeoff@gmail.com
|
|
|
06-25-2017, 05:03 AM
|
#8
|
Banned
Join Date: Dec 2016
Location: England
Posts: 2,432
|
Not sure what the octave and phase cut is with reaxcomp-but it can also be used.
Set to 1 band> (top frequency is what you sweep) > threshold @default(not used) ratio to 1 (0 compression) > small gain fader -150db(or less.)
The small gain fader now becomes your 'cut,or boost db level'
The top frequency can now be set to follow your fundamental lowest pitch I guess. =)
Not tested the phase reasponses tbh.
*EDIT* On closer inspection- reaxcomp has a resonance bump if anyone cares to measure that might be usefull..
Last edited by Bri1; 06-25-2017 at 06:16 AM.
|
|
|
06-25-2017, 09:14 AM
|
#9
|
Human being with feelings
Join Date: Mar 2016
Posts: 444
|
Quote:
Originally Posted by Mr. PC
Thanks.
I read that putting a minimum phase EQ on a reverb is fine because the reverb alters the signal anyhow, and so they're 2 different signals / no comb filtering.
Basically, comb filtering is the only problem that can be cause my phase issues, yes?
With a delay, if the delay is long enough, it also shouldn't be a problem, right? And if a delay is so short that it's getting into phase-problem territory, it could cause phase problems even without the EQ, right?
|
Whether it is a problem or not, depends on how you use it.
Example 1:
Reverb choice, adjustment etc -> EQ on reverb while listening to the mix -> tweak reverb -> go on with life
Example 2:
Reverb choice, adjustment etc -> HPF on reverb blindly or listening to reverb in isolation -> your mix may sound worse (or better) because HPF introduced phase shift to the reverb track and now it mixes differently with the dry signal
Same goes for delay or any other parallel FX.
Last edited by avocadomix; 06-25-2017 at 09:20 AM.
|
|
|
06-25-2017, 09:18 AM
|
#10
|
Human being with feelings
Join Date: Mar 2016
Posts: 444
|
Quote:
Originally Posted by bezusheist
does the comb filtering come from the further processing of the EQ'd track ?
i ask because using minimum phase filters in parallel does not cause comb filtering, so it must be from something else.
|
No, the comb filtering comes from applying minimum-phase EQ to a signal and then mixing in the unprocessed signal. This will inevitably cause weird phase-related issues that should be quite audible (unless you add a really negligible amount of the processed track, in which case it doesn't really matter how you process it).
|
|
|
06-25-2017, 09:59 AM
|
#11
|
Human being with feelings
Join Date: Nov 2010
Location: Mullet
Posts: 829
|
Quote:
Originally Posted by avocadomix
No, the comb filtering comes from applying minimum-phase EQ to a signal and then mixing in the unprocessed signal. This will inevitably cause weird phase-related issues that should be quite audible (unless you add a really negligible amount of the processed track, in which case it doesn't really matter how you process it).
|
But parallel filtering by itself does not cause comb filtering. It just alters the gain of the filter, or in the case of HP or LP filters it creates a shelf.
__________________
I like turtles
|
|
|
06-25-2017, 11:52 AM
|
#12
|
Human being with feelings
Join Date: Mar 2014
Location: germany
Posts: 196
|
parallel processing, okay. But more important is to think about EQing multi-mic´d sources. Drums, amps, any acoustic instrument with more than one mic on it.
The phase relations that took you so long to setup via micing, they can become completely screwed.
Sometimes you can use the phase shift to your advantage, though. I regularly mic kick drums with 2 mics, set up at equal distance from the batter head, and hi-pass one of them to remove mid-freq buildup
|
|
|
06-25-2017, 12:11 PM
|
#13
|
Human being with feelings
Join Date: Apr 2010
Location: Cloud 37
Posts: 1,071
|
Ok, I've been experimenting, and when I put a steep highpass on an instrument without it touching the fundamental (only cutting out room noise) I'm hearing absolutely no change in instrument tone. These are dry instruments recorded in isolation (mostly samples) and I'm not doing parallel processing. So in this case, it makes sense to just highpass as a rule, because it gets rid of noise (there's just a bit of noise, but better without it) and in theory should cause no problems, and I'm hearing no problems.
The smearing is a problem with parallel processing because you have 2 copies of the same signal, and one copy is having the phase of certain frequencies shifted. This will result in phase cancellation of certain frequencies. However if that signal is being processed by a reverb, doesn't the reverb alter the signal enough that it won't have any phase-interaction with the original?
|
|
|
06-25-2017, 03:00 PM
|
#14
|
Human being with feelings
Join Date: Mar 2016
Posts: 444
|
Quote:
Originally Posted by Mr. PC
However if that signal is being processed by a reverb, doesn't the reverb alter the signal enough that it won't have any phase-interaction with the original?
|
As long as 2 signals have common frequencies, they will interact when mixed, based on the phase. The question is whether this interaction is consistent (significant correlation) or not. Generally, for parallel FX, the shorter the predelay - the more correlation you get and the higher frequencies correlate. Algorithm consistency also plays a role here (as well as in doubled tracks, where the player takes the role of the algorithm in a sense). Talking about reverbs, especially in the lower frequencies, it is fairly easy to have a situation where your signal correlates with the delay or the reverb in a certain way. This is why when you change the predelay by a few ms, you may get a very different sound from the mix. If you fine-tune the reverb but then apply HPF to it without paying attention, it is easy to destroy the fine-tuning you had achieved earlier. However when you take your HPF high enough for the frequencies that remain to no longer have significant correlation with the dry signal, the phase interaction will no longer play a role. This may be one of the reasons why it is common to cut the reverb at about 500Hz with a HPF (that doesn't mean this is what everyone should do, just stating what many people do).
Last edited by avocadomix; 06-25-2017 at 03:30 PM.
|
|
|
06-25-2017, 03:13 PM
|
#15
|
Human being with feelings
Join Date: Mar 2016
Posts: 444
|
Quote:
Originally Posted by MrBongo
parallel processing, okay. But more important is to think about EQing multi-mic´d sources. Drums, amps, any acoustic instrument with more than one mic on it.
The phase relations that took you so long to setup via micing, they can become completely screwed.
|
This is true and a huge pitfall in multitrack recording.
Quote:
Sometimes you can use the phase shift to your advantage, though. I regularly mic kick drums with 2 mics, set up at equal distance from the batter head, and hi-pass one of them to remove mid-freq buildup
|
You can use an all-pass filter for that as well. Though this may be easier to achieve with a simple EQ scoop.
|
|
|
06-25-2017, 03:18 PM
|
#16
|
Human being with feelings
Join Date: Mar 2016
Posts: 444
|
Quote:
Originally Posted by bezusheist
But parallel filtering by itself does not cause comb filtering. It just alters the gain of the filter, or in the case of HP or LP filters it creates a shelf.
|
This is an interesting point. I think the result depends on the steepness of the filter. A 24db/octave filter or steeper IIRC should create a comb-filter effect. a 6db/octave filter may indeed create a shelf (not sure about it).
|
|
|
06-26-2017, 01:07 AM
|
#17
|
Human being with feelings
Join Date: Mar 2014
Location: germany
Posts: 196
|
Quote:
Originally Posted by avocadomix
You can use an all-pass filter for that as well. Though this may be easier to achieve with a simple EQ scoop.
|
definitely
I wanted to cut the low-end from one of the two mics anyway, so that came in handy.
ReaEQ got an all-pass filter, in case someone missed it
|
|
|
06-26-2017, 01:32 AM
|
#18
|
Human being with feelings
Join Date: Nov 2010
Location: Mullet
Posts: 829
|
Quote:
Originally Posted by avocadomix
This is an interesting point. I think the result depends on the steepness of the filter. A 24db/octave filter or steeper IIRC should create a comb-filter effect. a 6db/octave filter may indeed create a shelf (not sure about it).
|
yes, with 6 dB filter you get a shelf, with 12 dB you get a "pultec style" shelf with a "dip" before the boost, and with > 12 dB slopes you get a "notch", which is what people must be referring to by "comb filtering". i have always thought of comb filtering to be a series of these "notches", but i guess you could call this a one toothed comb...
here are the frequency plots of some min. phase HPF in parallel (50/50 dry/wet mix). (Fc = 100 Hz)
red = 6 dB
green = 12 dB
blue = 24 dB
orange = 48 dB
__________________
I like turtles
Last edited by bezusheist; 06-26-2017 at 01:40 AM.
|
|
|
06-26-2017, 04:32 AM
|
#19
|
Human being with feelings
Join Date: Mar 2009
Location: Dartmouth, Nova Scotia
Posts: 11,251
|
Quote:
Originally Posted by bezusheist
yes, with 6 dB filter you get a shelf, with 12 dB you get a "pultec style" shelf with a "dip" before the boost, and with > 12 dB slopes you get a "notch", which is what people must be referring to by "comb filtering". i have always thought of comb filtering to be a series of these "notches", but i guess you could call this a one toothed comb...
here are the frequency plots of some min. phase HPF in parallel (50/50 dry/wet mix). (Fc = 100 Hz)
red = 6 dB
green = 12 dB
blue = 24 dB
orange = 48 dB
|
Well, those are some possible curves for those general slopes.
However there is nothing that says, for instance, that 24 db/octave filters have a notch, check out Butterworth, Chebyshev, Linkwitz-Riley, Bessel, etc. to see some differences.
What you are referring to is called Q.
Each of the filters named have a Q associated with them (e.g. Butterworth Q = 0.707),
so while you have given some examples, there are no rules that say a given filter slope is tied to a particular Q, it's all in the design.
Comb filtering is another thing altogether, it comes into play when you mix 2 signals together.
__________________
To install you need the CSI Software and Support Files
For installation instructions and documentation see the Wiki
Donate -- via PayPal to waddingtongeoff@gmail.com
|
|
|
06-26-2017, 09:56 AM
|
#20
|
Human being with feelings
Join Date: Dec 2012
Posts: 7,294
|
Quote:
Originally Posted by Geoff Waddington
Comb filtering is another thing altogether, it comes into play when you mix 2 signals together.
|
That exactly what they were talking about!!!
I don't think there is such a thing as a one-toothed comb though. It's only a comb if the phase goes through 180 and comes all the way back around to odd multiples thereof. Even then, the two signals would have to be close enough in level for phase cancellation to make a noticeable difference in the final output. This is why we see one notch just below the cutoff frequency where the filtered signal is still pretty close to the unfiltered. Below that point, the unfiltered just dominates and won't be changed enough to notice.
|
|
|
06-26-2017, 04:28 PM
|
#21
|
Human being with feelings
Join Date: Nov 2009
Location: Montreal, Canada
Posts: 9,073
|
I have good news, you don't need to know this, and you also don't need to have your hi-pass automated to follow the fundamental. You just need to set it to where it sounds good. Anything you remove that doesn't make it sound worse, is fine to remove.
__________________
Slava Ukraini
|
|
|
06-26-2017, 11:41 PM
|
#22
|
Human being with feelings
Join Date: Jun 2009
Location: Croatia
Posts: 24,798
|
I wouldn't highpass anything at all if it absolutely isn't necessary, because highpass jacks up the phase of the signal at low frequencies, which can ruin transients, which is pretty important for some material (like multitracked drums for example).
Always look at the phase plot if your EQ/filter plugin offers it. For highpass, it goes all the way up at cutoff frequency. This is also why I would always use a shelft EQ rather than a highpass, if some rumble needs sorting out. But if there's no rumble in the signal, there's no need to highpass, at all.
|
|
|
06-27-2017, 03:13 AM
|
#23
|
Human being with feelings
Join Date: Nov 2010
Location: Mullet
Posts: 829
|
Quote:
Originally Posted by EvilDragon
Always look at the phase plot if your EQ/filter plugin offers it. For highpass, it goes all the way up at cutoff frequency.
|
it all depends on the filter.
and if you are referring to ReaEQ's phase plot, it is incorrect for some filters, HPF being one.
__________________
I like turtles
|
|
|
06-27-2017, 04:18 AM
|
#24
|
Human being with feelings
Join Date: Jun 2009
Location: Croatia
Posts: 24,798
|
Well, linear phase filters will have pre-ringing as their "thing" instead of jacking with the phase... Can't win, really.
I was referring to no particular plugin - ReaEQ is not the only one to supply the phase plot.
As for ReqEQ's phase plot, it would depend on which filter topology was used for those filters, I'm not entirely sure that the phase plot for HPF is wrong.
|
|
|
06-27-2017, 04:29 AM
|
#25
|
Human being with feelings
Join Date: Nov 2010
Location: Mullet
Posts: 829
|
Quote:
Originally Posted by EvilDragon
As for ReqEQ's phase plot, it would depend on which filter topology was used for those filters, I'm not entirely sure that the phase plot for HPF is wrong.
|
I checked in some other apps that show phase response and it seems that ReaEQ shows 2x the phase shift for the HPF.
For example, setting ReaEQ HPF with a butters Q (bandwidth approx 1.9) the corner Fc should be at 90 deg. but ReaEQ's phase plot shows it at 180 deg.
Edit: another way to confirm is apply HPF at x Hz to a sine wave at x Hz and look at the phase shift on a scope.
__________________
I like turtles
Last edited by bezusheist; 06-27-2017 at 04:38 AM.
|
|
|
06-27-2017, 07:52 AM
|
#26
|
Human being with feelings
Join Date: Nov 2009
Location: Montreal, Canada
Posts: 9,073
|
Quote:
Originally Posted by EvilDragon
I wouldn't highpass anything at all if it absolutely isn't necessary, because highpass jacks up the phase of the signal at low frequencies, which can ruin transients, which is pretty important for some material (like multitracked drums for example).
Always look at the phase plot if your EQ/filter plugin offers it. For highpass, it goes all the way up at cutoff frequency. This is also why I would always use a shelft EQ rather than a highpass, if some rumble needs sorting out. But if there's no rumble in the signal, there's no need to highpass, at all.
|
I find that things quickly get muddy at the low end, but I don't hear the phase disturbance you're referring to.
__________________
Slava Ukraini
|
|
|
06-27-2017, 08:02 AM
|
#27
|
Human being with feelings
Join Date: May 2009
Posts: 29,269
|
Quote:
Originally Posted by Sound asleep
I find that things quickly get muddy at the low end, but I don't hear the phase disturbance you're referring to.
|
If it does and you hear it, HPF it. If you don't hear it, don't HPF just because (exception being very low end rumble only visible on scope but that is still a quantified reason). I certainly don't have to HPF all the time, actually rarely and often get by just fine with a shelf if/when needed.
Personally, I think HPF is overrated and overused as a panacea.
__________________
Music is what feelings sound like.
Last edited by karbomusic; 06-27-2017 at 08:31 AM.
|
|
|
06-28-2017, 11:41 PM
|
#28
|
Human being with feelings
Join Date: Mar 2016
Posts: 444
|
Quote:
Originally Posted by bezusheist
yes, with 6 dB filter you get a shelf, with 12 dB you get a "pultec style" shelf with a "dip" before the boost, and with > 12 dB slopes you get a "notch", which is what people must be referring to by "comb filtering". i have always thought of comb filtering to be a series of these "notches", but i guess you could call this a one toothed comb...
here are the frequency plots of some min. phase HPF in parallel (50/50 dry/wet mix). (Fc = 100 Hz)
red = 6 dB
green = 12 dB
blue = 24 dB
orange = 48 dB
|
Thanks for taking measurement! Not sure why you are getting one notch at 48db/oct. The steeper filters are supposed to cause the phase plot to cross 360 degree line multiple times (iirc). This is why I called it a comb filtering effect. If I'm wrong then perhaps it's not a comb filtering but this is still an issue anyone trying to use minimum-phase filters in parallel needs to be aware of.
|
|
|
06-29-2017, 12:56 AM
|
#29
|
Human being with feelings
Join Date: Jul 2016
Posts: 33
|
Quote:
Originally Posted by avocadomix
Thanks for taking measurement! Not sure why you are getting one notch at 48db/oct. The steeper filters are supposed to cause the phase plot to cross 360 degree line multiple times (iirc). This is why I called it a comb filtering effect. If I'm wrong then perhaps it's not a comb filtering but this is still an issue anyone trying to use minimum-phase filters in parallel needs to be aware of.
|
What I see there is a kind of harmonics-creep.
You know how the more intense the rainbow, the more copies of the rainbow you see next to it? It seems that the more hardcore the Q and steepness, the more you start to see these extra wiggles appear.
You know when I think about it, a rainbow is probably quite similar to what we're doin here. I will think about this a little more.
|
|
|
06-29-2017, 08:02 AM
|
#30
|
Human being with feelings
Join Date: Dec 2012
Posts: 7,294
|
Quote:
Originally Posted by weblordpepe
What I see there is a kind of harmonics-creep.
...
You know when I think about it, a rainbow is probably quite similar to what we're doin here. I will think about this a little more.
|
Ummm... What?
It's really a lot more (kind of exactly) like the "3:1 Rule". 1 + (-1) = 0, but 1 + (-0.01) is pretty close to 1.
An all-pass that steep might show several notches because it's not also attenuating the frequencies that get the phase shift.
|
|
|
06-30-2017, 10:07 AM
|
#31
|
Human being with feelings
Join Date: Apr 2010
Location: Cloud 37
Posts: 1,071
|
Quote:
Originally Posted by avocadomix
As long as 2 signals have common frequencies, they will interact when mixed, based on the phase. The question is whether this interaction is consistent (significant correlation) or not. Generally, for parallel FX, the shorter the predelay - the more correlation you get and the higher frequencies correlate. Algorithm consistency also plays a role here (as well as in doubled tracks, where the player takes the role of the algorithm in a sense). Talking about reverbs, especially in the lower frequencies, it is fairly easy to have a situation where your signal correlates with the delay or the reverb in a certain way. This is why when you change the predelay by a few ms, you may get a very different sound from the mix. If you fine-tune the reverb but then apply HPF to it without paying attention, it is easy to destroy the fine-tuning you had achieved earlier. However when you take your HPF high enough for the frequencies that remain to no longer have significant correlation with the dry signal, the phase interaction will no longer play a role. This may be one of the reasons why it is common to cut the reverb at about 500Hz with a HPF (that doesn't mean this is what everyone should do, just stating what many people do).
|
Hmmm, this is very interesting. Yes, I almost always highpass by reverb around 500Hz.
So if the reverb isn't on a send (done in parallel). For example, if I put the reverb on the Master Track at 100% wet, this wouldn't be an issue, right? (at the loss of many things).
Quote:
Originally Posted by EvilDragon
I wouldn't highpass anything at all if it absolutely isn't necessary, because highpass jacks up the phase of the signal at low frequencies, which can ruin transients, which is pretty important for some material (like multitracked drums for example).
Always look at the phase plot if your EQ/filter plugin offers it. For highpass, it goes all the way up at cutoff frequency. This is also why I would always use a shelft EQ rather than a highpass, if some rumble needs sorting out. But if there's no rumble in the signal, there's no need to highpass, at all.
|
Hmmm, so is there no special EQ that doesn't harm transients? My ear isn't great for this, but when I used to use ReaEQ, I would avoid EQ unless needed, because things seemed to sound less-good. When I got Neutron, it seemed that highpassing did nothing to the sound other than remove noise.
That said, I don't use a lot of percussion (pizzicato more likely, though a slower attack isn't the end of the world for that).
Problem is, a lot of samples seem to have low-end noise (likely they have noise all over the spectrum, but you can't remove it without removing everything).
1 - So what I'm understanding is that phase-issues are only a problem for EQing in parallel.
2 - A parallel verb can possibly cause phase-issues in the low frequencies (which is why many high-pass the verbs).
3 - Highpass causes low-frequency phase to shift, ruining transients (only a problem for percussive / staccato sounds?)
I'm guessing the 'low frequency' would be the first harmonic here, since I'm high passing right before the fundamental; but what if there's some room between the high-pass and the fundamental, wouldn't that mean only noise is phase shifted? And I don't care how room noise interacts with other room noise. Maybe there's a rule for how much space is needed between the highpass and the fundamental?)
|
|
|
06-30-2017, 10:21 AM
|
#32
|
Human being with feelings
Join Date: Dec 2012
Posts: 7,294
|
Here's the thing. We can talk about the math all we want. We can use it to sort of predict the outcome of certain techniques, but the situation is extremely complex - chaotic in fact in the sense that the outcome is extremely sensitive to a number of fairly minor variables. So you'll have a whole lot of math to work out.
But most of us come equipped with a very finely tuned and very sensitive apparatus for measuring theses effects. You listen, and you know if it's working or not. The "science" can help maybe to figure out why it doesn't sound right and try to fix it, but in end, it just has to sound the way you want.
Part of what makes reverb what it is is the complex phase relationships between the multiple reflections. In fact, there is a type of reverb algorithm that's basically just a bunch of all-pass filters run in parallel.
A transient is only "ruined" if it doesn't sound right in the mix. It's kind of funny how people are willing to compress and distort their transients but get all weird about "phase smearing"...
|
|
|
06-30-2017, 12:11 PM
|
#33
|
Human being with feelings
Join Date: Jul 2007
Location: Jazz City
Posts: 5,074
|
32 posts in a thread about high pass filtering? Haha
Whatever sounds right. Watch out for phase anomalies in the context of the mix, especially with multiple drum mic tracks.
Start to mix on small speakers and get the mid range right, then switch to more low-end-accurate ones. You'll hear what's going on and correct what you think has to be corrected.
__________________
Windows 10x64 | AMD Ryzen 3700X | ATI FirePro 2100 | Marian Seraph AD2, 4.3.8 | Yamaha Steinberg MR816x
"If I can hear well, then everything I do is right" (Allen Sides)
|
|
|
06-30-2017, 12:14 PM
|
#34
|
Human being with feelings
Join Date: May 2009
Posts: 29,269
|
Quote:
Originally Posted by beingmf
32 posts in a thread about high pass filtering? Haha
|
I know right. I think everyone has to go down that rabbit hole and back out at least once, just to get it out of the way and back to the most important part, listening and using that as the ultimate decision maker. Proverbially, the only time you need to care about the math is when you are building one.
__________________
Music is what feelings sound like.
|
|
|
06-30-2017, 12:53 PM
|
#35
|
Human being with feelings
Join Date: Nov 2010
Location: Mullet
Posts: 829
|
i love high pass filters. i actually have a bumper sticker that says "i love high pass filters".
and a t shirt. well, the t shirt actually says "i love linear phase high pass filters"...on both sides...
i had another bumper sticker that said "gas, ass, or high pass", but my wife made me take it off the car
__________________
I like turtles
Last edited by bezusheist; 06-30-2017 at 01:03 PM.
|
|
|
06-30-2017, 02:07 PM
|
#36
|
Human being with feelings
Join Date: Mar 2016
Posts: 444
|
Quote:
Originally Posted by ashcat_lt
Here's the thing. We can talk about the math all we want. We can use it to sort of predict the outcome of certain techniques, but the situation is extremely complex - chaotic in fact in the sense that the outcome is extremely sensitive to a number of fairly minor variables. So you'll have a whole lot of math to work out.
But most of us come equipped with a very finely tuned and very sensitive apparatus for measuring theses effects. You listen, and you know if it's working or not. The "science" can help maybe to figure out why it doesn't sound right and try to fix it, but in end, it just has to sound the way you want.
|
I respect this approach but I think it only works for natural born mixers. Most of us are not this way and so our auditory system is easy to fool. In more precise terms, we are good at making conscious judgements on quality but only in a situation of immediate comparison. We are bad at spotting nuanced changes. Our unconscious perception, however, or in other words, the effect of sound quality on us, is more absolute and more sensitive to nuances. So it is very easy to create something you think sounds good, but in fact is muddled crap, unless you work in a way that minimizes the risk of being fooled, and that means employing best practices and understanding the physics, the electronics and the mathematics of the subject.
Last edited by avocadomix; 06-30-2017 at 02:14 PM.
|
|
|
06-30-2017, 02:29 PM
|
#37
|
Human being with feelings
Join Date: May 2009
Posts: 29,269
|
Quote:
Originally Posted by avocadomix
So it is very easy to create something you think sounds good, but in fact is muddled crap, unless you work in a way that minimizes the risk of being fooled, and that means employing best practices and understanding the physics, the electronics and the mathematics of the subject.
|
So "sounds good is good" is just a feel good statement. Yes, I mean that. But it doesn't require the math except for the engineering part, the listening part takes training ears which doesn't really need that math per se though can help get one out of a tough technical spot, or not blow up gear and so on. I'm not against the knowledge from an engineering perspective, far from it as I get annoyed if someone wants to be a recording 'engineer' and forgo the underlying principles that go with it.
Of course one of the reasons I'm even in this business in any form is because I had the tendency to be able to hear or zone in on nuance before I was ever doing any of this AND I had an interest/knack in the technical part, and the musical part, and the creative part.
__________________
Music is what feelings sound like.
Last edited by karbomusic; 06-30-2017 at 02:37 PM.
|
|
|
06-30-2017, 02:40 PM
|
#38
|
Human being with feelings
Join Date: Mar 2016
Posts: 444
|
Quote:
Originally Posted by karbomusic
So "sounds good is good" is just a feel good statement. Yes, I mean that. But it doesn't require the math except for the engineering part, the listening part takes training ears which doesn't really need that math per se though can help get one out of a tough technical spot, or not blow up gear and so on.
|
Even after years of practice, I keep making mistakes when not working carefully. The real question is always a comparative one because, again, our judgement only works well when comparing. So the question is not "does it sound good?" but rather "does it sound good next to the reference?".
Last edited by avocadomix; 06-30-2017 at 02:46 PM.
|
|
|
06-30-2017, 02:49 PM
|
#39
|
Human being with feelings
Join Date: May 2009
Posts: 29,269
|
Quote:
Originally Posted by avocadomix
Even after years of training, I keep making mistakes when not working carefully. The real question is always a comparative one because, again, our judgement only works well when comparing. So the question is not "does it sound good?" but rather "does it sound good next to the reference?".
|
I don't disagree per se. However, I'd suggest there are many who can create a damn fine mix, reference or no reference.
__________________
Music is what feelings sound like.
|
|
|
06-30-2017, 03:42 PM
|
#40
|
Human being with feelings
Join Date: Dec 2012
Posts: 7,294
|
I don't really think it's about "natural born" anything. Maybe some people are inclined more to noticing and remembering subtle details, but it really is more about practice. Listen listen listen. Mix mix mix. Listen some more and then mix some more. Things will come more quickly and easily and intuitively, but even the best occasionally end up loving the mix at the end of the night and then hating it the next morning every once in a while.
And no, I'm not at all trying to say that you shouldn't explore the science, learn the math, understand the underlying principles. That stuff is very important, but sometimes we end up just plain overthinking some of these things. The OP frankly seems to be near to paralysis from fear of phase smearing by this point, and even if it looks like a big deal on paper it almost never really becomes that big of an issue.
To jump off in a slightly different direction, consider...
About every piece of active gear that we use has at least two HPFs - at input and output - built in in the form of what are properly called "AC coupling", but often referred to as "DC blocking" capacitors. A full mixing board might have tens of these between the channel input and the master output. Your preamps have them, your interface has them, every meatspace audio box has them. These are generally chosen to cutoff way below the audio band, but they do add up.
Reaper, though, is DC coupled all the way through. Some plugins strip DC at input or output, but many really don't. Worse yet, some processes can create either DC offset or what amounts to or looks a lot like very low frequency waves that can steal headroom and make other processes act funny without actually being heard.
Last edited by ashcat_lt; 06-30-2017 at 03:53 PM.
|
|
|
Thread Tools |
|
Display Modes |
Linear Mode
|
Posting Rules
|
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts
HTML code is Off
|
|
|
All times are GMT -7. The time now is 12:58 PM.
|