You are here

Level and Frequency Masking?

For everything after the recording stage: hardware/software and how you use it.

Level and Frequency Masking?

Postby TomChimera » Mon Aug 10, 2020 12:19 am

Hi all,
I would be happy to understand this better and dig deeper with your help :)

I heard in a few occasions mixers say that cleaning frequency collision and buildup in frequencies, like kick & bass playing together, is actually only needed in order to free up energy so to be able to get more overall level in the end.
I wonder if it's true..

It seems to me that other than gaining level - freeing up buildup of energy, esthetically, it also reduces the resulted masking, in other words creates "space" and separation between sounds, and also the build up of energy in shared frequencies can sound like unwanted resonance peak..

If i'm wrong and cleaning colliding frequencies between sounds is indeed only relevant to getting level,
With today's standard of about -13 LUFS or in the future hopefully going lower,
it means that one can mix only with faders without eq, and the resulting level will be enough..
In this situation theoretically cleaning colliding frequencies will be an obsulete technique..
I find it hard to believe..


Second question:

I hear masking mentioned in 2 ways and I'm trying to understand these.
One, is while mixing, two sounds share the same Frequency.
And two, in mastering for example, where the bass is too loud and masks higher frequencies or similar, like "general" spectrum area masking..

Can you explain more about that

Thanks everybody :)
TomChimera
Poster
Posts: 22
Joined: Fri Jun 19, 2020 5:38 am

Re: Level and Frequency Masking?

Postby Tim Gillett » Mon Aug 10, 2020 6:35 am

TomChimera wrote:...I heard in a few occasions mixers say that cleaning frequency collision and buildup in frequencies, like kick & bass playing together, is actually only needed in order to free up energy so to be able to get more overall level in the end.
I wonder if it's true..

Yes that's true. Also, the bass can often have the largest say in the highest peak in a track as we tend to prefer more average bass than treble.

TomChimera wrote:It seems to me that other than gaining level - freeing up buildup of energy, esthetically, it also reduces the resulted masking, in other words creates "space" and separation between sounds, and also the build up of energy in shared frequencies can sound like unwanted resonance peak..

That's also true. Masking happens and has been well understood for many years. It's also true that the bass end can get very easily crowded out and one bass instrument can "compete" with another very quickly. Clean, enjoyable mixes - or orchestrations - are often ones where only one instrument, or one fundamental is heard in the lower spectrum. Nothing wrong with very deep bass notes as such though.

TomChimera wrote:If i'm wrong and cleaning colliding frequencies between sounds is indeed only relevant to getting level,
With today's standard of about -13 LUFS or in the future hopefully going lower,
it means that one can mix only with faders without eq, and the resulting level will be enough..
In this situation theoretically cleaning colliding frequencies will be an obsulete technique..
I find it hard to believe..

I think you were right the first time. Loudness averaging doesnt EQ. It only adjusts overall gain.


TomChimera wrote:Second question:

I hear masking mentioned in 2 ways and I'm trying to understand these.
One, is while mixing, two sounds share the same Frequency.
And two, in mastering for example, where the bass is too loud and masks higher frequencies or similar, like "general" spectrum area masking..

Can you explain more about that

Again both are true. In the second case, it may be not so much the bass masking the higher frequencies as just being too loud in the mix and being annoyingly loud to the listener. To help hear the higher tones more clearly we can turn up the gain but it now makes everything just louder, which doesnt solve the problem. The track needs to be remixed or perhaps just EQed.

Here's an example where even the lower speech tones (not even the deep bass tones from a double bass or kick drum) so overwhelm the trebles that it even compromises speech intelligibility for many listeners. So even the pro's can sometimes get it wrong. It's a judgement call and there's no substitute for listening carefully and comparing to well known acceptable recordings, and in the case of speech, comparing it to real life face-to-face voices which we encounter every day. It's sometimes said that we all know what real human voices sounds like. I'm not so sure.

https://youtu.be/9GsrDe96HUE
Tim Gillett
Frequent Poster (Level2)
Posts: 2212
Joined: Wed Jan 30, 2013 1:00 am
Location: Perth, Western Australia

Re: Level and Frequency Masking?

Postby TomChimera » Mon Aug 10, 2020 10:54 am

Hi Tim, Thanks for the informative answer :)

Tim Gillett wrote:Again both are true. In the second case, it may be not so much the bass masking the higher frequencies as just being too loud in the mix and being annoyingly loud to the listener. To help hear the higher tones more clearly we can turn up the gain but it now makes everything just louder, which doesnt solve the problem. The track needs to be remixed or perhaps just EQed.

Here's an example where even the lower speech tones (not even the deep bass tones from a double bass or kick drum) so overwhelm the trebles that it even compromises speech intelligibility for many listeners. So even the pro's can sometimes get it wrong. It's a judgement call and there's no substitute for listening carefully and comparing to well known acceptable recordings, and in the case of speech, comparing it to real life face-to-face voices which we encounter every day. It's sometimes said that we all know what real human voices sounds like. I'm not so sure.

https://youtu.be/9GsrDe96HUE

Great example, I was thinking about this just yesterday, I sat with friends, and a guy with rumbly low voice in this room the reflection made it so muddy, so it was harder to understand what he was saying, I think I sat in the room mode crest or something..

Also, Honestly I find it really hard to hear and identify such problems to the point I though it was my hearing..
The specific situation where an amount of energy in one frequency area, makes other frequencies un-intelligible.
For example, if there is a muddy bass (comparable to the voice in you example) of course that If I would apply compression, it reacts automatically to and reduces lower energy more than high, and also makes prominent transients, so it makes it clearer.
But if I will try to solve it only with EQ, I find it very hard! Any tips?
And I have seen, mostly mastering engineers do magic in this..

Tim Gillett wrote:It's sometimes said that we all know what real human voices sounds like. I'm not so sure.
I imagine calibrating a studio with a familiar voice recording, like a sentence one have heard all their childhood like mom saying "wash the dishes!" This is true reference, it sits deeply in the reptilian brain :)

Tim Gillett wrote:I think you were right the first time. Loudness averaging doesnt EQ. It only adjusts overall gain.

I'll try to explain better what I meant.
If level were the only concern (disregarding masking),
Lets say I'm mixing a track without using EQ to clean competing frequencies.
The resulting level or loudness of the mix is -21 LUFS and highest peak at -1.
Theoretically, With the current and upcoming loudness standards, this can be good as is.
While if I were to clean those competing areas I would be at -13 LUFS...

It reminds me I noticed Dave Darlington mixing jazz wasn't using EQ for competing areas at all (https://www.youtube.com/watch?v=Tmpd-9ukSr8)
TomChimera
Poster
Posts: 22
Joined: Fri Jun 19, 2020 5:38 am

Re: Level and Frequency Masking?

Postby blinddrew » Mon Aug 10, 2020 11:07 am

I think the loudness is a bit of a red-herring here, masking can happen at any volume, it's really just about more than one source occupying the same frequency range. And since it's a question of dynamics and frequency we have a host of dynamic- and frequency-based tools to help us out. Plus panning of course.
Say you've got a vocal and guitar, both centrally panned, they might well be fighting with each other across their frequency ranges. So you could start by band-limiting the guitar so that you're not getting any low frequency mud and you're leaving the very high frequency for the vocals. you then might apply a bit of a scoop to the upper mids so that that the vocal dominates this part of the spectrum for intelligibility. You could then stick a compressor on the guitar and side-chain it to the vocal so that it's ducked whenever the vocal part kicks in.
If you've got a bass that needs a bit of controlling and the compressor is then adversely affecting stuff further up the frequency range then you could use a multi-band compressor so that it's only working on the low frequency component.
It's really all just about carving space for each instrument to sit in the mix and understanding that, in doing so, stuff might sound 'wrong' in isolation. And that's ok.
User avatar
blinddrew
Jedi Poster
Posts: 11179
Joined: Sun Jul 05, 2015 12:00 am
Location: York
Ignore the post count, I have no idea what I'm doing...

Re: Level and Frequency Masking?

Postby CS70 » Mon Aug 10, 2020 11:29 am

"Freeing" headroom is one (but not the only one) of the consequences of cutting low frequencies, simply because low frequencies require lot of energy to be reproduce... long wavelength, lots of air to move. Cutting high frequencies doesn't really give back much headroom at all, because the energy in these freqs simply isn't much.

In broad terms, ultimately "headroom" means the maximum energy that you can pass thru (and put out of) an energy-processing system (in our case, an audio playback system) before it starts to mangle the signal, aka distort. So if you lower the low frequencies a little, you'll reduce the total energy required to emit the signal quite a bit, and have more left for the rest.

"Masking" is about the simple fact that what we perceive is the rate of energy change in a specific frequency band. When two sounds share a similar amount of energy in the same frequency band, these will add up and there's more chance that the summed energy stays nearer to the same amount - i.e. chance that the rate of change is lower - so it can be harder for our ears to distinguish the sound.

Masking may happen in any band obviously, but our ears are most sensitive to the midrange, so tiny energy changes over time in that band are easier to distinguish for us. Which is probably one of the reasons that having too much midrange energy is a common mistake of mixing beginners: you can still make up all the stuff, so it's doesn't sound like a problem.
User avatar
CS70
Jedi Poster
Posts: 6030
Joined: Mon Nov 26, 2012 1:00 am
Location: Oslo, Norway
Silver Spoon - Check out our latest video and the FB page

Re: Level and Frequency Masking?

Postby TomChimera » Mon Aug 10, 2020 11:47 am

blinddrew wrote:I think the loudness is a bit of a red-herring here, masking can happen at any volume

Thanks! this is what I thought and this clarifies it.

I was very surprised when I heard mixers confidently saying that when instruments competing over the frequency range, masking is not a problem, but it's only a problem of "wasted" energy..

Oh, just found a good example. There is English subtitles
Just the sentence on the kick and bass and RMS..
https://youtu.be/7tel8HvbuPs?t=2524

I have heard this few times, also in longer explanations people disregarding masking completely.. So it made me wonder, hence the question :)
TomChimera
Poster
Posts: 22
Joined: Fri Jun 19, 2020 5:38 am

Re: Level and Frequency Masking?

Postby Hugh Robjohns » Mon Aug 10, 2020 12:23 pm

TomChimera wrote:I heard in a few occasions mixers say that cleaning frequency collision and buildup in frequencies, like kick & bass playing together, is actually only needed in order to free up energy so to be able to get more overall level in the end.
I wonder if it's true..

I wouldn't say it's only needed for that purpose, but that is certainly one very good reason that it was done routinely in many music genres where squashing the life out of a track was deemed necessary.

It seems to me that ... it also reduces the resulted masking, in other words creates "space" and separation between sounds...

Yes, this is another very good reason -- giving individual instruments the spectral space to sit alongside everything else without trampling over each other.

It's often the case that when people mix only in stereo, the spatial separation that comes from panning instruments apart allows very similar sounds to remain separate and distinct, yet when the track is summed to mono they get surprised at the resulting mush. This is especially common with double-tracked guitars etc.

So carving out frequency regions of one source to make way for another is a fundamental requirement of crafting a good mix of tonally complex music.
User avatar
Hugh Robjohns
Moderator
Posts: 28310
Joined: Fri Jul 25, 2003 12:00 am
Location: Worcestershire, UK
Technical Editor, Sound On Sound

Re: Level and Frequency Masking?

Postby TomChimera » Mon Aug 10, 2020 12:28 pm

CS70 wrote:"Freeing" headroom is one (but not the only one) of the consequences of cutting low frequencies, simply because low frequencies require lot of energy to be reproduce... long wavelength, lots of air to move. Cutting high frequencies doesn't really give back much headroom at all, because the energy in these freqs simply isn't much.

In broad terms, ultimately "headroom" means the maximum energy that you can pass thru (and put out of) an energy-processing system (in our case, an audio playback system) before it starts to mangle the signal, aka distort. So if you lower the low frequencies a little, you'll reduce the total energy required to emit the signal quite a bit, and have more left for the rest.

"Masking" is about the simple fact that what we perceive is the rate of energy change in a specific frequency band. When two sounds share a similar amount of energy in the same frequency band, these will add up and there's more chance that the summed energy stays nearer to the same amount - i.e. chance that the rate of change is lower - so it can be harder for our ears to distinguish the sound.

Masking may happen in any band obviously, but our ears are most sensitive to the midrange, so tiny energy changes over time in that band are easier to distinguish for us. Which is probably one of the reasons that having too much midrange energy is a common mistake of mixing beginners: you can still make up all the stuff, so it's doesn't sound like a problem.

Very interesting, thank you!

CS70 wrote:chance that the rate of change is lower - so it can be harder for our ears to distinguish the sound.

If I understand correctly,

This could explain why when two sounds occupy the same range like vocal and guitar, it creates for me the illusion of standing waves or resonant peaks, even if they were not present in them soloed, because the accumulated frequencies lowers the rate of change and makes it sound like "standing" wave, as it is really technically longer at the frequency.

In a way.. clearing up competeing frequencies in a mix, which is shortening the rate of change, is similar to shortening the impulse response of a sound reproduction system... because the brain feeds on time domain transient cues, and can more clearly recognize sounds, so it make them more intelligible...

So masking is basically about revealing micro time cues for the brain?
TomChimera
Poster
Posts: 22
Joined: Fri Jun 19, 2020 5:38 am

Re: Level and Frequency Masking?

Postby Hugh Robjohns » Mon Aug 10, 2020 12:51 pm

CS70 wrote:In broad terms, ultimately "headroom" means the maximum energy that you can pass thru (and put out of) an energy-processing system...

No. That's the Peak Level.

In simple terms, Headroom is the space between the peak level and, technically, the normal operating level or, in more practical terms, the average signal level.

The higher the average signal level, the less the available headroom.

"Masking" is about the simple fact that what we perceive is the rate of energy change in a specific frequency band. When two sounds share a similar amount of energy in the same frequency band, these will add up and there's more chance that the summed energy stays nearer to the same amount - i.e. chance that the rate of change is lower - so it can be harder for our ears to distinguish the sound.

Frequency or spectral masking is a very complex subject, but fundamentally it's about not being able to hear a signal of one frequency in the presence of a louder signal at a different by nearby frequency, largely due to the way the cochlea's basilar membrane changes shape in an effort to gain greater pitch resolution of the louder signal at the expense of reduced sensitivity to other signal pitches nearby.

A very intriguing and sobering demonstration I often run for my students is to set up a constant 3kHz tone at a very low but audible level from a single speaker. And then mix over that tone a much louder slowly upward sweeping tone starting at 40Hz or so, and continuing to 15kHz or higher.

Some experimentation with levels may be necessary, and the tones and speaker have to have negligible distortion, obviously. But what happens is that the 3kHz tone is perceived as being switched off as the sweep frequency approaches, and a surprising time later, gets switched back on again as the sweep tone gets well past it...

Some students find it quite distressing to learn that there ears are telling them lies... but this is the reason why we can mangle signals so horrendously with lossy data codecs like MP3 and not notice!

There is also the phenomena of temporal masking where the first arriving sound is ignored in favour of a louder subsequent sound. Something else which the more elaborate lossy codecs take advantage of.
User avatar
Hugh Robjohns
Moderator
Posts: 28310
Joined: Fri Jul 25, 2003 12:00 am
Location: Worcestershire, UK
Technical Editor, Sound On Sound

Re: Level and Frequency Masking?

Postby TomChimera » Mon Aug 10, 2020 1:27 pm

Hugh Robjohns wrote:A very intriguing and sobering demonstration I often run for my students is to set up a constant 3kHz tone at a very low but audible level from a single speaker. And then mix over that tone a much louder slowly upward sweeping tone starting at 40Hz or so, and continuing to 15kHz or higher.

Hi Hugh, I wish I were in this lesson,
This sounds distressing just reading about it, I'm going to try this today.

Referring to what CS70 said, about rate of change.
Do you think time, length of sounds is also part of masking?

Another example, lets say there's drums and guitar, the kick drum lingers long, in around 50 hz, the guitar up in the mid frequencies can sound masked by it, and shortening the kick makes the guitar pop up... I wonder why it happens
TomChimera
Poster
Posts: 22
Joined: Fri Jun 19, 2020 5:38 am

Re: Level and Frequency Masking?

Postby Tim Gillett » Mon Aug 10, 2020 1:31 pm

Thanks for that explanation Hugh and the description of the experiment which I will try to set up for my own interest because I'm curious to find out what this test looks like on a spectral analyzer and whether the 3 kHz tone temporarily disappears on the analyzer just as it seems to do so to our hearing.
Tim Gillett
Frequent Poster (Level2)
Posts: 2212
Joined: Wed Jan 30, 2013 1:00 am
Location: Perth, Western Australia

Re: Level and Frequency Masking?

Postby CS70 » Mon Aug 10, 2020 2:01 pm

Hugh Robjohns wrote:
CS70 wrote:In broad terms, ultimately "headroom" means the maximum energy that you can pass thru (and put out of) an energy-processing system...

No. That's the Peak Level.

"Peak level" is a property of the signal passed thru a system, not the system passing it.
The headroom of a system can be defined simply as the level it can process given that there is no signal present, and I don't think it's particularly misleading, in broad terms: people for example speak of headroom as the distance between seat and roof of a car, or between head and roof, interchangeably.

Frequency or spectral masking is a very complex subject

Yeah it is - and the fact that masking occurs over bands rather than individual freqs is why I mentioned them. ;-) However, perhaps the best is to simply refer to chapter 10 from Gelfand's book for details, as anything short of that can be nitpicked to no end, if one so wants to spend his time.
User avatar
CS70
Jedi Poster
Posts: 6030
Joined: Mon Nov 26, 2012 1:00 am
Location: Oslo, Norway
Silver Spoon - Check out our latest video and the FB page

Re: Level and Frequency Masking?

Postby TomChimera » Mon Aug 10, 2020 2:11 pm

CS70 wrote:perhaps the best is to simply refer to chapter 10 from Gelfand's book for details, as anything short of that can be nitpicked to no end, if one so wants to spend his time.

What is the book's name please? :)
TomChimera
Poster
Posts: 22
Joined: Fri Jun 19, 2020 5:38 am

Re: Level and Frequency Masking?

Postby CS70 » Mon Aug 10, 2020 2:45 pm

TomChimera wrote:
CS70 wrote:perhaps the best is to simply refer to chapter 10 from Gelfand's book for details, as anything short of that can be nitpicked to no end, if one so wants to spend his time.

What is the book's name please? :)

I have two on the subject - one is "Psychological Acoustics" by Erik Young (Springer) and the one I was referring to is "Hearing, an introduction to psychological and physiological acostuics" by Stanley Gefland (Dekker).. both are good intros I'd say.
User avatar
CS70
Jedi Poster
Posts: 6030
Joined: Mon Nov 26, 2012 1:00 am
Location: Oslo, Norway
Silver Spoon - Check out our latest video and the FB page

Re: Level and Frequency Masking?

Postby TomChimera » Mon Aug 10, 2020 3:16 pm

CS70 wrote:I have two on the subject - one is "Psychological Acoustics" by Erik Young (Springer) and the one I was referring to is "Hearing, an introduction to psychological and physiological acostuics" by Stanley Gefland (Dekker).. both are good intros I'd say.

Thanks!
For anybody interested Gefland's book is available for free here:

It is a very interesting book, but I'm not convinced that Russian site has the publisher's authority to provide their copyrighted material to all and sundry for free... So, in respect of the Forum Rule: Copyright material must not be posted unless it is owned either by you or by this web site I've removed the link -- HR.
Edit: Oh Sorry! it was just the first result in google and I didn't think about it :thumbup:

And do you mean this part - "Physiological Acoustics" - Eric Young - Pages 429-457,
In here? https://link.springer.com/referencework ... 87-30425-0
TomChimera
Poster
Posts: 22
Joined: Fri Jun 19, 2020 5:38 am

Next