You are here

Level and Frequency Masking?

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

Re: Level and Frequency Masking?

Postby Hugh Robjohns » Mon Aug 10, 2020 3:16 pm

Tim Gillett wrote: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...

It looks like a static spike at 3kHz and a moving finger of the sweep tone ranging left to right.

...and whether the 3 kHz tone temporarily disappears on the analyzer just as it seems to do so to our hearing.

It partly depends on the quality and resolution of your FFT-based spectrum analyser, but no, the 3kHz signal doesn't disappear, the display just gets overwritten by the passing sweep 'finger' and re-emerges immediately the finger has passed. So the narrower the 'finger' the more obvious that point is!

I usually run this demo in Adobe Audition, and put the frequency analysis display on screen while it's running just to prove that the tone remains present and unchanged even though it's inaudible. I run the FFT at the highest resolution (FFT size = 65536) with the 'area' display mode and Blackman-Harris windowing.

The reason that the display doesn't emulate what we actually hear is that an FFT analyser doesn't work with physical active and dynamic organic structures which divide frequency analysis into 'critical bands' in the way our sense of hearing does. Instead, it just analyses what's really there, using maths... :D :ugeek:
User avatar
Hugh Robjohns
Moderator
Posts: 28544
Joined: Fri Jul 25, 2003 12:00 am
Location: Worcestershire, UK
Technical Editor, Sound On Sound

Re: Level and Frequency Masking?

Postby Hugh Robjohns » Mon Aug 10, 2020 3:23 pm

CS70 wrote:"Peak level" is a property of the signal passed thru a system, not the system passing it.

Well... if you're going to be even more pedantic about it than me, :ugeek: :lol: you're right and I should have used the term 'Maximum Permitted Level' which is the official EBU definition... although I think most people would have known what I meant. :think:

Nevertheless, the point I was trying to make is that the term 'Headroom' has a very specific meaning in the context of all audio systems, which is the difference between the 'Maximum Permitted Level' and the 'Alignment Level' (or nominal operating level in 'old speak').

The point where the system "starts to mangle the signal, aka distort. ..." is when you've run out of headroom and reached the maximum permitted level.

Shared understanding only comes from using the correct terms in the correct ways within the correct contexts... And especially so in technical matters... So I felt I should make the point for the avoidance of further confusion. Not trying to be a smartass, just trying to clarify.
User avatar
Hugh Robjohns
Moderator
Posts: 28544
Joined: Fri Jul 25, 2003 12:00 am
Location: Worcestershire, UK
Technical Editor, Sound On Sound

Re: Level and Frequency Masking?

Postby Kwackman » Mon Aug 10, 2020 3:31 pm

Hugh Robjohns wrote:It is not "the maximum energy that you can pass thru (and put out of) an energy-processing system...".
That sounds like a binge session involving many packets of Oaties and a willing guinea pig.... :bouncy:
User avatar
Kwackman
Frequent Poster
Posts: 1892
Joined: Thu Nov 07, 2002 1:00 am
Location: Belfast
Cubase, guitars.

Re: Level and Frequency Masking?

Postby CS70 » Mon Aug 10, 2020 6:32 pm

Hugh Robjohns wrote:Not trying to be a smartass, just trying to clarify.

Haha neither did I :D I normally use the term with the same meaning as you - referring to an existing signal - but when I wrote the above I was truly thinking in system terms. For example, at the beginning of my practical audio journey, I found headroom described as the "hidden buffer over of the 0 level of a meter" (or suchlike, from people used to analogue meters).. Like in "18 dB of headroom over 0".. which is a kind of system definition, as it does not require a specific signal, but it means that, from the zero point as shown by say a VU meter, you have additional 18dB available at any time - it's kinda irrelevant to mention a signal since you have a zero reference. Only afterwards I met the "headroom as a function of the signal" definition (which is useful to understand gain staging, for example). The two are correlated of course, and it's kinda obvious how once you think about it, so - unless I rigorously define the exact meaning of the word - I tend to use them interchangeably ("in broad terms").

Anyways, definitely better with one clarification more than one less - the goal is to reply to the OP question.
User avatar
CS70
Jedi Poster
Posts: 6177
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 CS70 » Mon Aug 10, 2020 6:36 pm

TomChimera wrote:And do you mean this part - "Physiological Acoustics" - Eric Young - Pages 429-457,
In here? https://link.springer.com/referencework ... 87-30425-0

I did!

The three meters between the library and my desk are obviously too many for my verbal memory :D
User avatar
CS70
Jedi Poster
Posts: 6177
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 CS70 » Mon Aug 10, 2020 6:38 pm

Kwackman wrote:
Hugh Robjohns wrote:It is not "the maximum energy that you can pass thru (and put out of) an energy-processing system...".
That sounds like a binge session involving many packets of Oaties and a willing guinea pig.... :bouncy:

And when you start learning about general systems theory in our times, you can't but think that Iain M. Banks 's gotta be involved. :D
User avatar
CS70
Jedi Poster
Posts: 6177
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 » Thu Aug 13, 2020 2:23 pm

Hi all, thanks again for the answers, it really made it much clearer for me.

I'm reading through Gefald chapter on masking and thought some people might also be interested in understading masking further,
so I made a very simplistic summary of a few points.
Please tell me if anything I wrote is wrong :)


1. Low freq mask higher freq more than high freq mask low freq.

2. Low freq mask mostly the higher freq that are closest to them. ("..vicinity of the masker...")

- Practical potential use:
A. You have a rich range instrument like a piano or bass, and there is a resonance or some accumulation of energy somewhere in the mids.
High pass it bellow the problem area temporarily and you will be able to hear more clearly the problem.
B. You have a bass rich mix, or any mix for that matter, high pass it to around 100hz (NS10 style) to reveal problems in higher freq.

3. Obviously masking increases as the intensity of the masker is raised compares to the maskee (the signal being masked).

4. "Upward spread of masking"
Masking spread, changes depends on the relative intensity.
For example, a masker sound at 1000hz at 40 db and a maskee at 20 db - the spread of masking will be symmetrical at this level.
The spread will shift higher as the masker raises.
So the 1000hz masker at 100db will mask much more the higher freq with little change to lower freq.

(See diagram, low left square, Gelfand's book, page 188)

Meaning:
The higher the level of a sound -> ability to hear frequencies above it decreases..
Or specifically, When the bass volume raises -> ability to hear higher freq decreases.
As the bass masking reach is wide, (See next #5 - the lower the masker's freq the wider is it's reach of masking higher)

And the parts closest to it will suffer the most (#2), in a bass rich mix, the low mids will be the part you will hear the least and will be most illusive to your ears.
If your mix is bass rich, it becomes even more crucial to check it high passed.

(I wish I knew this sooner...)

5. The lower the freq of the Masker the more spread it have upward.
As you probably noticed, a resonance peak in the mids, have a masking spread quite close to it.
While low freq resonance will have much more reach upward in it's masking.

(this is shown in the same diagram, at Gelfand's book, page 188)
*Not sure I can post it here due to copyright material

6. When the maskee and the masker are close in frequencies they can create combination tones that produces beats and it might be hard to determine if the result is beats or masking.

7. "Critical bandwidth of the auditory filter",
When the maskee is at the center frequency of a white noise masker.
and the white noise is filtered using low pass and high pass, there is a certain width which making the resulting filter wider beyond it, will cease to increase masking.
(although it would make the white noise louder)
This is the limit of the critical bandwidth of the ear's filter.

8. "Off-frequency listening"
The "Critical bandwidth of the auditory filter" in the ear is not always "focused" or "listen" in the area of the filter center.

Interesting phenomena which might suggest a counter-intuitive technique to experiment with:

So lets say there is some masking happen around 400hz.
While the ear's auditory filter is centered around that frequency (on-signal band) it might be hard to hear what is masked, obviously.. :)
if we increase some frequency somewhere else nearby (flanking or off-frequency bands) maybe at ~450hz, shifting our ear filter bandwidth there,
we might be able to better hear what's going on at 400hz.

Or you want to judge and focus better in what is going on in an area of the mix,
increase some other frequency nearby...
shifting the ear filter center a bit might result in less masking in the original intended area of focus..

*Not sure this is true but i'm trying to interept it to practical advice.

9. This also relates to "comodulation masking release"
- "Comodulation masking release reveals that the auditory system is able to capitalize upon information provided across critical band filters,
although a cohesive model explaining CMR is not yet apparent.
One type of explanation suggests that the information provided by the off-signal band helps the subject
know when the troughs or “dips” occur in the modulating noise.
Listening for the signal during these dips would result in a lower threshold (less masking)..."
- "Notice that the masked threshold does not improve if the on-signal
and off-signal noise bands are not comodulated"


Which means that increasing off-frequency bands will reveal sounds (reduce masking) only if there is modulation in the sound, static sounds will not work the same..

10. Masking Overshoot
When two sounds, a maskee and masker are played together the masking is highest.
The more the maskee is delayed compared to the masker, the less masking is happening.

*This relates and explains something I really love and find very interesting,
that as music is played live, transient information is rarely played together at a similar resolution and accuracy like digital DAWs are playing, of course...
When we create music in a DAW placing everything exactly on the beat, it is rather unrealistic, and creates masking.



That's it for now, I'm just in the middle of the chapter and will do the experiments mentioned above...

I hope this may benefit you
TomChimera
Poster
Posts: 28
Joined: Fri Jun 19, 2020 5:38 am

Re: Level and Frequency Masking?

Postby CS70 » Thu Aug 13, 2020 6:25 pm

Haven't read all the summary (I shall) but that was a cool result. Much better to provide sources rather (or a link to something more substantial) than a post sometimes, that cannot but present things in really rough terms (or with just the conclusions).

Now do yourself a favor and go buy a copy of the book on Amazon - just an electronic one if you must - dodgy internet links aren't really kosher, and Stanely's work deserves better than being lifted for free ;-)
User avatar
CS70
Jedi Poster
Posts: 6177
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 » Thu Aug 13, 2020 10:17 pm

CS70 wrote:Now do yourself a favor and go buy a copy of the book on Amazon - just an electronic one if you must - dodgy internet links aren't really kosher, and Stanely's work deserves better than being lifted for free ;-)

Yes I'm buying it second hand from ebay, I believe this is also respectful to Stanely's work, and the planet too..
TomChimera
Poster
Posts: 28
Joined: Fri Jun 19, 2020 5:38 am

Re: Level and Frequency Masking?

Postby CS70 » Fri Aug 14, 2020 9:36 am

TomChimera wrote:
CS70 wrote:Now do yourself a favor and go buy a copy of the book on Amazon - just an electronic one if you must - dodgy internet links aren't really kosher, and Stanely's work deserves better than being lifted for free ;-)

Yes I'm buying it second hand from ebay, I believe this is also respectful to Stanely's work, and the planet too..

Attaboy!
User avatar
CS70
Jedi Poster
Posts: 6177
Joined: Mon Nov 26, 2012 1:00 am
Location: Oslo, Norway
Silver Spoon - Check out our latest video and the FB page

Previous