>15 kHz? (+Master filtering)

Have an urge to learn, or a calling to teach? Want to share some useful Youtube videos? Do it here!
RobC
Posts: 1166
Joined: 10 Mar 2018

Post 23 Sep 2019

Be it mixing, or mastering, doesn't matter. General audio engineering.

I was confused so far, thinking every engineer does low pass filtering. This might have come from when I read tons of articles about mastering for vinyl.

For modern media, it's definitely not a requirement to apply LPF. However, I kind of worry about frequencies above 15 kHz, since during adulthood, supposedly, our ears get less sensitive in those regions. I think, my top is about 17 kHz, with a sine tone test, above that, not much I could hear.

Now, if I don't hear anything after a certain point, then I'm afraid I could mess up the equalization.

Generally, most mastering engineers say 'f*cking don't'. This is nice and all, but I see people asking this question on forums - but there's never an answer to it.

I learned an interesting thing, though: they say, not to touch the final mix with filters. Honestly, I feel a linear phase filter doesn't wreck the mix - although of course, I can just filter mix channels individually. Still, when it comes to the mix-down, some low frequency rumble can still happen, especially if you add master effects (most commonly filter sweeps), or take the final mix-down and do some slicing and creatively play around with the waveform (such as glitching/juggling effects).

All in all, question is:

- what to do with frequencies one might not hear?
- can master-filtering (meaning engineering, not as an effect) really be safely avoided?

User avatar
Loque
Posts: 6000
Joined: 28 Dec 2015

Post 23 Sep 2019

In theory a 22khz wave influences a a 1khz wave. There is a point, where you cannot here it if you remove everything about that point (like in your example 17khz), but i guess, you can hear it, if you add a 6db LPF at 17khz, because it changes the overall sound and due to phasing or filtering the overall sound changes too - it does not matter to use linear phase or not. The results of what you are doing always depend.

From my perspective, i rather try not to use a EQ or filter in the master chain. I'd rather try to get my bass clean, the frequencies separated and the top end not too noise. How successfully i am with that is another story :-D
:reason: 11, Win10 64Bit.

User avatar
selig
Moderator
Posts: 8003
Joined: 15 Jan 2015

Post 23 Sep 2019

As for high frequencies, think of octaves/semitones instead of frequencies. The difference between 17k and 20 kHz is only about 3 semitones. You would have to do some seriously sharp EQing to accidentally boost/cut at 20 kHz when working with frequencies above 10 kHz. How often do you use a super sharp EQ curve on frequencies above 10 kHz?

Remember there is only one octave between 10k and 20 kHz, and most common EQ curves you would use, especially a high shelf, as SO gentle you would never be affecting frequencies you cannot hear when working with those you can hear.

Also note that for most sounds except pure white noise, each higher octave's harmonics are softer than the lower octave. Meaning, anything that is present around 20 kHz is softer than what's around 10 kHz.

But even using the extreme white noise example, if you boost at 12 kHz with a high shelf it should be noted you are also boosting to some degree far BELOW 12 kHz in addition to ABOVE. In other words, you'll hear it, no problem, and you can assume you're not adding too much of anything you can't hear as long as you don't add too much of anything you CAN hear!

Hopefully this makes sense - and to "+1" what Loque says.
Selig Audio, LLC

RobC
Posts: 1166
Joined: 10 Mar 2018

Post 24 Sep 2019

Loque wrote:
23 Sep 2019
In theory a 22khz wave influences a a 1khz wave. There is a point, where you cannot here it if you remove everything about that point (like in your example 17khz), but i guess, you can hear it, if you add a 6db LPF at 17khz, because it changes the overall sound and due to phasing or filtering the overall sound changes too - it does not matter to use linear phase or not. The results of what you are doing always depend.

From my perspective, i rather try not to use a EQ or filter in the master chain. I'd rather try to get my bass clean, the frequencies separated and the top end not too noise. How successfully i am with that is another story :-D
Okay, I had no clue about that (definitely not consciously) - or at least, most manual, or tutorials never warn about this. The ones I found, that is.

I'd like to avoid using anything at all after mixdown ~ but like I said, if I still tinker around with the final waveform, and I do a lot, then problems can happen. Non-zero-crossing cuts can create pops/clicks - which a lot of times, I dig, however looking at the spectrum, those can in some cases add a momentary sub rumble, and ultra frequencies.
selig wrote:
23 Sep 2019
As for high frequencies, think of octaves/semitones instead of frequencies. The difference between 17k and 20 kHz is only about 3 semitones. You would have to do some seriously sharp EQing to accidentally boost/cut at 20 kHz when working with frequencies above 10 kHz. How often do you use a super sharp EQ curve on frequencies above 10 kHz?

Remember there is only one octave between 10k and 20 kHz, and most common EQ curves you would use, especially a high shelf, as SO gentle you would never be affecting frequencies you cannot hear when working with those you can hear.

Also note that for most sounds except pure white noise, each higher octave's harmonics are softer than the lower octave. Meaning, anything that is present around 20 kHz is softer than what's around 10 kHz.

But even using the extreme white noise example, if you boost at 12 kHz with a high shelf it should be noted you are also boosting to some degree far BELOW 12 kHz in addition to ABOVE. In other words, you'll hear it, no problem, and you can assume you're not adding too much of anything you can't hear as long as you don't add too much of anything you CAN hear!

Hopefully this makes sense - and to "+1" what Loque says.
I made a topic here, cause good old FB doesn't care much for notifying me...

I kind of fell for what they say about linear phase EQs that they would be 'surgically precise'. They are neat for splitting a signal into multiple bands, but still far from perfect.
As for references, well ~ the high cut was more about what I said here, too: tons of vinyl mastering articles (if I remember right); as well as studying human hearing system.

Good to know, all in all, that high cut does more bad than good for digital audio. I never liked the idea of cutting that half top octave, as something just didn't feel right - especially if I created a multi-band split for creating a graphical EQ, cause the 10th band was just a half one.

And the only sharp EQ is pretty much that ~10-20 kHz band. I thought of doing 1/3rd octave bands at some point (about 30 ~ where the last band, with an imaginary perfect EQ, wouldn't be audible), but I doubt I'll do that much and precise shaping anyway.

Seems all clear now.

There's one more danger area: what about when it comes to down-sampling from anything above 44100 Hz sample rate? It's said, some filter at 22050 Hz is used to reduce aliasing.

jlgrimes
Posts: 358
Joined: 06 Jun 2017

Post 24 Sep 2019

RobC wrote:
23 Sep 2019
Be it mixing, or mastering, doesn't matter. General audio engineering.

I was confused so far, thinking every engineer does low pass filtering. This might have come from when I read tons of articles about mastering for vinyl.

For modern media, it's definitely not a requirement to apply LPF. However, I kind of worry about frequencies above 15 kHz, since during adulthood, supposedly, our ears get less sensitive in those regions. I think, my top is about 17 kHz, with a sine tone test, above that, not much I could hear.

Now, if I don't hear anything after a certain point, then I'm afraid I could mess up the equalization.

Generally, most mastering engineers say 'f*cking don't'. This is nice and all, but I see people asking this question on forums - but there's never an answer to it.

I learned an interesting thing, though: they say, not to touch the final mix with filters. Honestly, I feel a linear phase filter doesn't wreck the mix - although of course, I can just filter mix channels individually. Still, when it comes to the mix-down, some low frequency rumble can still happen, especially if you add master effects (most commonly filter sweeps), or take the final mix-down and do some slicing and creatively play around with the waveform (such as glitching/juggling effects).

All in all, question is:

- what to do with frequencies one might not hear?
- can master-filtering (meaning engineering, not as an effect) really be safely avoided?
I would use a lowpass filter for mastering. Even if you can't hear it, someone else probably can.



What I do sometimes though is use a high shelving filter to turn down the highs some as this can smoothen out the high end which I do hear.

RobC
Posts: 1166
Joined: 10 Mar 2018

Post 24 Sep 2019

jlgrimes wrote:
24 Sep 2019
I would use a lowpass filter for mastering. Even if you can't hear it, someone else probably can.



What I do sometimes though is use a high shelving filter to turn down the highs some as this can smoothen out the high end which I do hear.
It seems a filter does damage elsewhere on the spectrum, though.

User avatar
Boombastix
Posts: 824
Joined: 18 May 2018
Location: Bay Area, CA

Post 24 Sep 2019

Well, there is a bit of a phase shift below the cut off frequency, so you would have to listen. If you use VSTs directly in the digital domain, there can be a lot of very high frequencies that analog recorded stuff never really gets into. Another problem is of course ear fatigue in general when mixing or the listening room not being well treated or speakers are throwing things off. But usually the high frequencies that have the energy and is picked up well by the ear/brain is lower than 17kHz.

When in doubt, why not just try one of the AI smart EQs. Try them on a material that you feel is good and representative, see of the AI EQ keeps it flat at 10kHz and above. Then try your stuff and see what AI response you get. If don't have an AI EQ then just run a match EQ against a ref track and see what it does in the high bands.

I don't find standard spectrum analysers being the best here since they move so fast and the range is 0-70dB, you can of course try to freeze the spectrum and zoom in, so maybe, but still a bit tricky for me to get use of it to check >10k frequencies.

Another way can be to use a good reference track and high pass it at 10k or 12k and make a A/B ref comparison. Once you only listed to this domain it is easier to hear any difference than listening to everything at once. This a pretty useful method for any referencing work and splitting it into 4 bands and do A/B for each band.
10% off at Waves with link: https://www.waves.com/r/6gh2b0
Disclaimer - I get 10% as well.

User avatar
aeox
Posts: 2289
Joined: 23 Feb 2017

Post 24 Sep 2019

I never liked the sound of high cutting my master section, but I do it all the time for individual instruments if they need it.

Something I always do is linear phase low cutting in the master section. To me, it "tightens" up the bass to my ears. I guess it's because it's removing everything below 30hz so there is more room in the mix for the bass you can actually feel?
Random songs/ideas go here:
https://soundcloud.com/ae-ox/polar/s-Ocxtj

User avatar
Loque
Posts: 6000
Joined: 28 Dec 2015

Post 24 Sep 2019

I found the videos of Fabfilter very good. It's worth watching, especially with linear phase stuff
https://www.fabfilter.com/products/pro- ... er-plug-in
:reason: 11, Win10 64Bit.

RobC
Posts: 1166
Joined: 10 Mar 2018

Post 25 Sep 2019

Boombastix wrote:
24 Sep 2019
Well, there is a bit of a phase shift below the cut off frequency, so you would have to listen. If you use VSTs directly in the digital domain, there can be a lot of very high frequencies that analog recorded stuff never really gets into. Another problem is of course ear fatigue in general when mixing or the listening room not being well treated or speakers are throwing things off. But usually the high frequencies that have the energy and is picked up well by the ear/brain is lower than 17kHz.

When in doubt, why not just try one of the AI smart EQs. Try them on a material that you feel is good and representative, see of the AI EQ keeps it flat at 10kHz and above. Then try your stuff and see what AI response you get. If don't have an AI EQ then just run a match EQ against a ref track and see what it does in the high bands.

I don't find standard spectrum analysers being the best here since they move so fast and the range is 0-70dB, you can of course try to freeze the spectrum and zoom in, so maybe, but still a bit tricky for me to get use of it to check >10k frequencies.

Another way can be to use a good reference track and high pass it at 10k or 12k and make a A/B ref comparison. Once you only listed to this domain it is easier to hear any difference than listening to everything at once. This a pretty useful method for any referencing work and splitting it into 4 bands and do A/B for each band.
To be completely honest, whenever I did a pretty much brick wall cut with a flexible FFT based filter (by GoldWave), it visibly cleaned up waveforms, while I heard no difference in the audible regions. The pre-echoing and similar artifacts were pretty much non-existent, too.

For analysis, I often rendered regions on the spectrum to see and hear what's going on.

Of course, I will be less than likely to be able to do that in case of oversampling, when building tools in Reaktor. I will definitely need filtering to avoid aliasing.
aeox wrote:
24 Sep 2019
I never liked the sound of high cutting my master section, but I do it all the time for individual instruments if they need it.

Something I always do is linear phase low cutting in the master section. To me, it "tightens" up the bass to my ears. I guess it's because it's removing everything below 30hz so there is more room in the mix for the bass you can actually feel?
What type of high cut filter did you try? I never liked analog style ones. However, the linear phase ones with brick wall cuts can be really interesting for sound design.

I think you noticed when it comes to linear phase, that it can create pre-echoing, especially if you cut into an audible frequency. It can add some low range smoothness, so it's not necessarily always a bad thing.

During design however, when I look for enhancing clicks, knocks, thumps, etc, I definitely prefer the analog-style EQs. Meanwhile I like linear phase for gently and smoothly balancing things out (that's where I really dig brick wall style multi band splitting).
Loque wrote:
24 Sep 2019
I found the videos of Fabfilter very good. It's worth watching, especially with linear phase stuff
https://www.fabfilter.com/products/pro- ... er-plug-in
Yep, those helped me a lot to decide which EQ, and where to use. Many articles, tutorials, studies failed to explain the differences that simply, or didn't show what's going on 'under the hood'.

The practices where they make room on the spectrum, are interesting. I used to do these, too, but to be honest, I never liked how they changed the tonality of sounds. Even if the mix is cleaner in the end (I don't necessarily enjoy that more than what's 'muddy'). I rather looks for ways to avoid that as much as possible. Of course, that will touch the sound in other ways, so I'm not saying it's better whatever things I try, either. But most of these traditional methods don't exactly think outside the box IMO. That's why I messed around with phase earlier for example.
Lately, I thought of creating a 3D effect processor, but that again is kind of like imitating traditional stereo '3D' recordings. It's impressive when moving around 1 sound, but it's nothing special when there are multiple sounds (whether in one place, or moving around). That's why I also keep juggling between 3D, dual mono effects, simple panning, etc.

User avatar
selig
Moderator
Posts: 8003
Joined: 15 Jan 2015

Post 25 Sep 2019

aeox wrote:
24 Sep 2019
I never liked the sound of high cutting my master section, but I do it all the time for individual instruments if they need it.

Something I always do is linear phase low cutting in the master section. To me, it "tightens" up the bass to my ears. I guess it's because it's removing everything below 30hz so there is more room in the mix for the bass you can actually feel?
IF you're not already cleaning up your low end during mixing (highly suggested), then it's possible you'll need to clean it up during mastering - but ONLY if there's something there to clean up. The idea you MUST add filtering to every master is what most would argue against, since every master does not have the need for filtering.
In short, only clean what is dirty…
Selig Audio, LLC

User avatar
selig
Moderator
Posts: 8003
Joined: 15 Jan 2015

Post 25 Sep 2019

RobC wrote:
25 Sep 2019
To be completely honest, whenever I did a pretty much brick wall cut with a flexible FFT based filter (by GoldWave), it visibly cleaned up waveforms, while I heard no difference in the audible regions.
Since no music fan buys "waveforms", and few would have an FFT, to quote Duke Ellington: "if it sounds good, it IS good."
I tend to not add processing I cannot hear…
:)
Selig Audio, LLC

User avatar
aeox
Posts: 2289
Joined: 23 Feb 2017

Post 25 Sep 2019

selig wrote:
25 Sep 2019
aeox wrote:
24 Sep 2019
I never liked the sound of high cutting my master section, but I do it all the time for individual instruments if they need it.

Something I always do is linear phase low cutting in the master section. To me, it "tightens" up the bass to my ears. I guess it's because it's removing everything below 30hz so there is more room in the mix for the bass you can actually feel?
IF you're not already cleaning up your low end during mixing (highly suggested), then it's possible you'll need to clean it up during mastering - but ONLY if there's something there to clean up. The idea you MUST add filtering to every master is what most would argue against, since every master does not have the need for filtering.
In short, only clean what is dirty…
I do that during mixing first but then when I do it during mastering again it always sounds better to me. So maybe I've just got too much low frequencies? I don't know, I just tried it and thought it sounded good and went with it, but I'm always eager to learn!
Random songs/ideas go here:
https://soundcloud.com/ae-ox/polar/s-Ocxtj

User avatar
selig
Moderator
Posts: 8003
Joined: 15 Jan 2015

Post 25 Sep 2019

aeox wrote:
25 Sep 2019
selig wrote:
25 Sep 2019


IF you're not already cleaning up your low end during mixing (highly suggested), then it's possible you'll need to clean it up during mastering - but ONLY if there's something there to clean up. The idea you MUST add filtering to every master is what most would argue against, since every master does not have the need for filtering.
In short, only clean what is dirty…
I do that during mixing first but then when I do it during mastering again it always sounds better to me. So maybe I've just got too much low frequencies? I don't know, I just tried it and thought it sounded good and went with it, but I'm always eager to learn!
If it sounds better when you do it in mastering, why wait for mastering to do it? Why not "fix" the problems at the mix level? The better I got at mixing, the less I needed to do when mastering. Most of what I do when mastering is only concerning mastering multiple songs so they sound more "fluid" in sequence. When mastering a single song, I typically only do a little limiting 3-6 dB max) and sometimes stereo widening (love Ozone Elements for this). Most of the work to get the master to sound as I want is done at the mix stage. :)
Selig Audio, LLC

User avatar
aeox
Posts: 2289
Joined: 23 Feb 2017

Post 25 Sep 2019

selig wrote:
25 Sep 2019
aeox wrote:
25 Sep 2019


I do that during mixing first but then when I do it during mastering again it always sounds better to me. So maybe I've just got too much low frequencies? I don't know, I just tried it and thought it sounded good and went with it, but I'm always eager to learn!
If it sounds better when you do it in mastering, why wait for mastering to do it? Why not "fix" the problems at the mix level? The better I got at mixing, the less I needed to do when mastering. Most of what I do when mastering is only concerning mastering multiple songs so they sound more "fluid" in sequence. When mastering a single song, I typically only do a little limiting 3-6 dB max) and sometimes stereo widening (love Ozone Elements for this). Most of the work to get the master to sound as I want is done at the mix stage. :)
I tried that but was using more instances of the plugin and was causing me to max out DSP :D


I mean, I could just put one instance of it on the sub channel or eq/adjust levels, etc but it still doesn't sound as good to me that way; with my limited mixing knowledge this was the easiest solution I found by trial and error.

Will keep trying to achieve the sound without putting on the master section, I appreciate your input :)
Random songs/ideas go here:
https://soundcloud.com/ae-ox/polar/s-Ocxtj

RobC
Posts: 1166
Joined: 10 Mar 2018

Post 25 Sep 2019

selig wrote:
25 Sep 2019
RobC wrote:
25 Sep 2019
To be completely honest, whenever I did a pretty much brick wall cut with a flexible FFT based filter (by GoldWave), it visibly cleaned up waveforms, while I heard no difference in the audible regions.
Since no music fan buys "waveforms", and few would have an FFT, to quote Duke Ellington: "if it sounds good, it IS good."
I tend to not add processing I cannot hear…
:)
I meant, when I used to throw away what little was above 15 kHz, with GoldWave's Spectrum Filter, I didn't notice any damage in lower regions. And I didn't miss what I removed above 15 kHz either.

You once said, something like that the mixdown is one big wave[form] - which it is. I'm not sure, but I guess all Linear Phase / FIR filters use FFT processing.

I just used to cut out what I didn't hear. But we all seem to do at least a cut below 20 Hz - yet we can't hear that either.

RobC
Posts: 1166
Joined: 10 Mar 2018

Post 25 Sep 2019

aeox wrote:
25 Sep 2019
I do that during mixing first but then when I do it during mastering again it always sounds better to me. So maybe I've just got too much low frequencies?
What Linear Phase filter do you use?

I mean, for me, GoldWave's spectrum filter either does a clean cut, or if I add unnecessary extreme settings, and let it go crazy with artifacts, then I hear some gentle pre-echoing / attack ~ which can be fun.

IIR filters are the other possibility, which don't make a perfect cut, as they do a bit of cutting and boosting around the cut point due to some phasing issue or whatever (I'm too tired). I can't stand that though. But it really does make a kick sound harder (but also less subby).

User avatar
Boombastix
Posts: 824
Joined: 18 May 2018
Location: Bay Area, CA

Post 25 Sep 2019

About linear phase on bass/kicks. This is what you need to see:

Pay attention around the 1min mark.
10% off at Waves with link: https://www.waves.com/r/6gh2b0
Disclaimer - I get 10% as well.

User avatar
Boombastix
Posts: 824
Joined: 18 May 2018
Location: Bay Area, CA

Post 25 Sep 2019

selig wrote:
25 Sep 2019
If it sounds better when you do it in mastering, why wait for mastering to do it? Why not "fix" the problems at the mix level?
There should be more focus on this aspect. The biggest favor you can do if you have a problem with a mix is: change the/a sound. Especially kick/snare. I have encountered numerous mixing "problems", believing a particular sound wasn't good, and tried to tweak. Sometimes it had to do with the drums not being good enough. It is funny how many sounds start to sound much better in the mix if the drums are perfect. But sometimes you just have to scrap the sound and change it to something else, EQ can only so much. If you're stuck with an acoustic recording you can of course try to salvage, but it can be a lot of work. Sometimes I layer a synth sound, played like the acoustic, just to get the tonal aspect better, if needs to be "rescued".
10% off at Waves with link: https://www.waves.com/r/6gh2b0
Disclaimer - I get 10% as well.

User avatar
aeox
Posts: 2289
Joined: 23 Feb 2017

Post 25 Sep 2019

RobC wrote:
25 Sep 2019
aeox wrote:
25 Sep 2019
I do that during mixing first but then when I do it during mastering again it always sounds better to me. So maybe I've just got too much low frequencies?
What Linear Phase filter do you use?

I mean, for me, GoldWave's spectrum filter either does a clean cut, or if I add unnecessary extreme settings, and let it go crazy with artifacts, then I hear some gentle pre-echoing / attack ~ which can be fun.

IIR filters are the other possibility, which don't make a perfect cut, as they do a bit of cutting and boosting around the cut point due to some phasing issue or whatever (I'm too tired). I can't stand that though. But it really does make a kick sound harder (but also less subby).
Waves Linear Phase EQ (low band)
Random songs/ideas go here:
https://soundcloud.com/ae-ox/polar/s-Ocxtj

RobC
Posts: 1166
Joined: 10 Mar 2018

Post 26 Sep 2019

Boombastix wrote:
25 Sep 2019
About linear phase on bass/kicks. This is what you need to see:

Pay attention around the 1min mark.
I just watched that video (again), too, yesterday. Agreed that Linear Phase is no good for dialing in those extreme boosts. Like I say, I just do brick wall multi-band splits. Then I only do gentle adjustments with the manually created graphical EQ.
aeox wrote:
25 Sep 2019
Waves Linear Phase EQ (low band)
Yeah, either you might experience some pre-echoing, or you do a clean cut, which also cleans your bass frequencies up, and the rest of your dynamic processing might be doing the thing. Possibly (but not necessarily) with the interaction of kick and bass instruments (since you said that on their own, the effect doesn't really happen).

RobC
Posts: 1166
Joined: 10 Mar 2018

Post 26 Sep 2019

selig wrote:
25 Sep 2019
If it sounds better when you do it in mastering, why wait for mastering to do it? Why not "fix" the problems at the mix level? The better I got at mixing, the less I needed to do when mastering. Most of what I do when mastering is only concerning mastering multiple songs so they sound more "fluid" in sequence. When mastering a single song, I typically only do a little limiting 3-6 dB max) and sometimes stereo widening (love Ozone Elements for this). Most of the work to get the master to sound as I want is done at the mix stage. :)
Speaking of Ozone Elements, since it's for free ATM, and it seems to have good reputation, I grabbed it, too. The EQ seems nice - though I'm sure you're more experienced with testing it. Any opinion?

So far, looks like it won't replace GoldWave's precise spectrum EQ for creating Linear Phase multi-band splits (for me). At least I didn't see a way to use it as a splitter (/frequency crossover). Though I just quickly looked at it's surgical mode.

But it generally seems like a decent tool.

User avatar
Chizmata
Posts: 439
Joined: 21 Dec 2015

Post 27 Sep 2019

even if you stop hearing above 15 (or even 5) khz, you will still be able to hear phase effects introduced by inaudible high frequencies.

User avatar
selig
Moderator
Posts: 8003
Joined: 15 Jan 2015

Post 27 Sep 2019

Chizmata wrote:
27 Sep 2019
even if you stop hearing above 15 (or even 5) khz, you will still be able to hear phase effects introduced by inaudible high frequencies.
Not sure I follow. What examples can you give that demonstrate what you're describing?
Selig Audio, LLC

User avatar
Chizmata
Posts: 439
Joined: 21 Dec 2015

Post 27 Sep 2019

if you lose your hearing of high frequencies, you can still hear the indirect effects of those frequencies on the stuff you can still hear. just like sounds above 20khz leave their traces in the audible region for everyone because its still physics, no matter if you hear it or not.

  • Information
  • Who is online

    Users browsing this forum: CommonCrawl [Bot] and 0 guests