• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Device for Equal Loudness/Fletcher Munson Curve? Do Any Speakers Adapt to This?

Aerith Gainsborough

Addicted to Fun and Learning
Joined
May 4, 2020
Messages
853
Likes
1,280
If I hade a library and only listened to that I would probably scan the whole thing and normalize everything to the highest possible integrated LUFS level > and then calibrate to this level
Yes, pretty much all my sources are offline media because I dislike streaming services.
I wonder if normalization can be done on the fly w/o actually changing the files. Probably via replay gain?

When you generate pink noise in Audacity for -14 LUFS you can do as @dasdoingHow did you determine that the test signal has 66 dB SPL (C) on your system?

And why did you set digital volume to -12 dBFS?

Calibrated UMIK + REW at the listening position (I listen nearfield ~1m from the speakers).
Digital volume is at -12dB due to Dirac Live, it's the headroom it reserves itself for doing the equalization. I can venture a bit beyond that but I have to keep an eye on it, depending on the piece I might run into digital clipping.
 

dasdoing

Major Contributor
Joined
May 20, 2020
Messages
4,209
Likes
2,674
Location
Salvador-Bahia-Brasil
I wonder if normalization can be done on the fly w/o actually changing the files. Probably via replay gain?

normalization can't be done on the fly since it needs a loudness average of a song, or in case of an album the average of that.
I never got into the solutions too much, but I am pretty sure many exist that don't change the original file. about replay gain, Spotify abandoned it in favor of a LUFS solution. There isn't realy a perfect solution, even LUFS has it's issues.
 

Aerith Gainsborough

Addicted to Fun and Learning
Joined
May 4, 2020
Messages
853
Likes
1,280
normalization can't be done on the fly since it needs a loudness average of a song, or in case of an album the average of that.
I never got into the solutions too much, but I am pretty sure many exist that don't change the original file. about replay gain, Spotify abandoned it in favor of a LUFS solution. There isn't realy a perfect solution, even LUFS has it's issues.
Foobar can write replay gain data into the tag area of the files and convert it into the LUFS metric.
Clip.png


The files need to be scanned before, but actual sound data is not altered.
The range is rather large: This soundtrack goes from files with -8.4 all the way down to -20.9.
That's a lot of fiddling with the values during listening.
 

dasdoing

Major Contributor
Joined
May 20, 2020
Messages
4,209
Likes
2,674
Location
Salvador-Bahia-Brasil
Foobar can write replay gain data into the tag area of the files and convert it into the LUFS metric.View attachment 101480

The files need to be scanned before, but actual sound data is not altered.
The range is rather large: This soundtrack goes from files with -8.4 all the way down to -20.9.
That's a lot of fiddling with the values during listening.

you shouldn't scan albums on a per track basis. they are mastered to combine allready. prety sure the software has an option to scan albums
 

Aerith Gainsborough

Addicted to Fun and Learning
Joined
May 4, 2020
Messages
853
Likes
1,280
you shouldn't scan albums on a per track basis. they are mastered to combine allready. prety sure the software has an option to scan albums
True in most cases but I don't think it applies to the SE OST above.
Seems more like a fan made collection (I think I found it on Youtube but I don't remember).
Some tracks are REALLY loud compared to others, forcing me to turn them down.

Yes Foobar can do both: track and Album.

The downside: if I scan everything, I'll have to re-upload ~120GB into the cloud. Ugh. <.<
I think I'll wait until I have the RME and can actually apply the correction.
Thanks for all the input though, I think I now see a way that could work w/o having to fiddle with the controls on every second track.
 

dasdoing

Major Contributor
Joined
May 20, 2020
Messages
4,209
Likes
2,674
Location
Salvador-Bahia-Brasil
True in most cases but I don't think it applies to the SE OST above.
Seems more like a fan made collection (I think I found it on Youtube but I don't remember).
Some tracks are REALLY loud compared to others, forcing me to turn them down.

Yes Foobar can do both: track and Album.

The downside: if I scan everything, I'll have to re-upload ~120GB into the cloud. Ugh. <.<
I think I'll wait until I have the RME and can actually apply the correction.
Thanks for all the input though, I think I now see a way that could work w/o having to fiddle with the controls on every second track.

good luck
 

Aerith Gainsborough

Addicted to Fun and Learning
Joined
May 4, 2020
Messages
853
Likes
1,280
Yup!
That's the string I added to the playlist view.
 
Joined
Feb 8, 2019
Messages
22
Likes
20
Location
Australia
The effect was unsettling. Why? Well, we studio engineers (who were not psycho-acoustic experts) surmised that humans live with the equal loudness effect from birth and unconsciously ‘compensate’ for it quite effectively. We get used to quiet audio having less bass (and, to a smaller extent, less top) and that somehow seems normal and what the ear/brain expects.
I've come to this thread months too late, but just in case: I've been arguing for years that applying loudness compensation EQ is a fundamentally misconceived idea.
I review gear for a living. Back when this compensation came in on home theatre receivers -- Audyssey in Marantz and Denon -- and it was in Onkyo/Integra for a while -- and Yamaha's own system ... I found the sound dissatisfying. It took me a while to work out why. Initially the receivers were switching this on by default, and I just went with it. Once I realised what was going on, I of course took to switching it off and the sound returned to how it should properly be. Over the next few years I'd criticise the fact these things were on by default. Eventually -- and I doubt that it was my influence, since I merely write reviews in Australian magazines -- they took to having these off by default. Thank goodness.

(Question: I don't normally read other reviewers -- to much groupthink in this game -- but I did make an effort to see who else was complaining about this adjustment. I couldn't find anyone. Why?)

I published a fuller coverage of this in Australian HI-FI magazine (Sept/Oct 2016, v.47#05). I'll post this in a sec. But tl;dr: your hearing expects a certain frequency balance at a given volume level. Fiddle with the frequency balance to change it to something your hearing would expect at a high level, when playing at a low level, it's going to sound strange and artificial.
 
Joined
Feb 8, 2019
Messages
22
Likes
20
Location
Australia
Avoid the Loudness

When I was a youngster and avid reader of Australian HI-FI, nearly every amplifier review contained, it seemed, a mini rant on the presence of a particular button fitted to the amp. Now, many years later, I find myself writing a mini rant on what is essentially the same subject in our sister magazine, Sound+Image, in most of my reviews of home theatre receivers.

That control was the 'Loudness' control. These days that control is a great deal more sophisticated, thanks to digital signal processing technology, and goes under various names depending on the brand, such as Audyssey Dynamic EQ, Dolby Volume, YPAO Volume and so on. These do the intended job better than the 'Loudness' control and overcome some of the inevitable issues it had. But that kind of misses the point: all of these things, and the original 'Loudness' button as well, were and are fundamentally misconceived because they are based on a misunderstanding of how our hearing works. Our hearing, note, not our ears.

It all started in 1933 when the Fletcher–Munson curves were published. This came out of research that indicated that human hearing was not linear, but the sense of loudness varied by frequency. Over the decades further experimental work has been conducted, so there are significant variations between the original Fletcher–Munson curves and the equal loudness contours specified in ISO 226.2003, but the details don't matter too much to us here.

What does matter is what those curves mean. They show that if you play a 1kHz tone at 60dB SPL, a 30 hertz tone will have to be more than 80dB SPL to seem like it's the same loudness. Taking things to their end points, extreme treble and bass can evade detection by the human ear at much higher levels than mid frequencies. A healthy young person can hear 3kHz or 4kHz at as low a level as -8dBSPL. That is, at an even quieter level than that originally specified as the threshold of hearing. But to hear 100 hertz it has to be at least 35dBSPL, and to hear 30 hertz it has to be at least 60dB SPL.

Likewise, at 15kHz a healthy young person's hearing is 15dB to 20dB less sensitive than it is at 3kHz.

Our equipment makers seeing those things think: ah, we can shape the sound to get address that 'problem'. We can boost the bass and treble when the volume control is low so that balance is 'restored'. (And yes, those are all 'scare' quotes.) That's what a 'Loudness' control does. It boosts bass and treble. The better ones would boost them more at lower settings of the volume knob, and not at all at higher settings.

There was at least one obvious problem with that. The amplifier didn't know how big your room was, how far away from the speakers that you were sitting, how efficient your speakers were. And it turns out, the varying sensitivity of the ear depends not only on frequency but on volume level. At 100dB SPL (on the newer equal loudness curves) 30 hertz at 100dBSPL sounds close as dammit in level to 1kHz at 100dBSPL. So not only were 'Loudness' controls fundamentally misconceived, as I'm arguing here, they wouldn't even work to properly address the 'problem' they for which they were designed, unless your system happened to sound exactly as loud as their 'corrections' were designed for.

Modern DSP-based 'correction' processes (at least in theory) deal with this, because they are set up using data gathered by the receiver in the automatic room calibration process. So the processor knows how loud the sound is where you are sitting and can apply the inverse of the appropriate equal loudness contour.

MISCONCEPTIONS

But here's the thing, you and I and everyone else does not hear low level sound as deficient in those frequencies to which our ears are less sensitive. Our ears are only part of our hearing mechanism. The signals from them are fed into an enormously sophisticated organic signal processor, the human brain. (Not just humans: other animals would have similar mechanisms).

This OSP does astonishing things. By comparing subtle phase differences between the left and right ear signals it can determine the direction from which the sound is coming (which is why bass isn't directional -- the wave length is too large for the distance between your ears to allow the accurate determination of the gap between the crests of the wave).

And, it turns out, our hearing mechanisms have a powerful EQ processor. Here I make a bold claim: the world around you will continue to sound subjectively the same as you age. The upper reaches of your hearing sensitivity rolls off as the years accumulate, undeniably and measurably so. As a result in crowded and noisy environments you will find it harder to distinguish what someone is saying to you because those high frequencies are used, if available, by the brain to better localise sound and to allow focus on a particular source. But all you will know is that you find it harder to understand what someone is saying. It will still seem like it sounds the same as it always did.

This works a bit like the auto-white-balance in your eyes. Back when we used to use film cameras, taking a photo under incandescent lighting would result in an orange hue, which was invisible to your eyes when you were taking it. Because your seeing mechanism takes the video feed from your eyeballs and processes it to look 'right', based to a degree on averaging over time.

And so it is with your ears. Test it yourself. Take a little cotton wool, loosely ball it up and put it in your ear canal. Gently, at the very surface, in just far enough so it won't easily fall out, but so that you can readily take it out. Leave it there for a couple of days -- actually, replace it a couple of times along the way. Note the dullness. But after a while sounds will no longer seem dull. They'll sound kind of normal. But you'll have greater trouble understanding what's being said to you, of picking out one voice among many while conversing in a crowd.

Then take out the cotton wool, and suddenly your world will be full of the tinkle of glassware, birds tweeting outside the window, the oddly bright voices of those around you. For a little while. But soon enough everything will sound pretty much the same as normal as your hearing recalibrates to the newly bright signal.

That is, I suspect, why some hearing impaired people detest their hearing aids. Everything sounds harsh and bright on first use, and only by leaving it in for a reasonable period without interruption can the hearing mechanism adjust to the tonal balance of the new signal and make it sound right.

FUNDAMENTAL MISCONCEPTION

So the fundamental misconception of all those processors, starting with the 'Loudness' control, is that the undeniable differences in hearing sensitivity at different frequencies and volumes requires some kind of redress.

It doesn't, because our hearing already provides this. For sure, the treble component of the music is reduced in level relative to the midrange when the music is being played at background levels. But our hearing expects it to sound lower in level. If doesn't, the music is going to sound harsh. The same is the case with respect to bass, although in this case it tends to make the deeper bass sound disconnected from the rest of the sound.

Your hearing expects the world to have a certain frequency balance at any given volume level. Alter that and it's going to sound wrong.

GRADUAL IMPROVEMENTS

There was a period a few years ago when many brands not only had 'Dynamic' 'Volume' processors, but switched them on by default. That was a dark period indeed for sound quality and I wonder how many people found their equipment less satisfying because of this issue.

Now, receivers mostly still have the processors, but switch them off by default. Thank goodness, although better yet would be to eliminate them altogether.

And if, perchance, you are using a home theatre receiver, make sure that any such processing is disabled. Look under the Audio or Audyssey setup menu, or consult the receiver's manual.

When you first switch it off, if you're used to it, your system may initially sound a little dull. Just persist for while until the true character of music has a chance to assert itself.
 
Joined
Feb 8, 2019
Messages
22
Likes
20
Location
Australia
for a day or 2
Interesting. Now you've gotten me to thinking. If you read the fuller piece (posted after your comment) you'll see my entire argument. But could it be that one can have a two-track hearing process? Because, remember, the rest of the world is not loudness compensated.

If you live in front of your audio system (actually, I kind of do live in front of one of mine!) and don't experience other audio environments, then your brain's signal processor will probably adjust. Good point.

I'm just not sure that it's an adjustment that I want.

But perhaps one's personal signal processor can switch between modes.
 

RayDunzl

Grand Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
13,201
Likes
16,983
Location
Riverview FL
Here I make a bold claim: the world around you will continue to sound subjectively the same as you age. The upper reaches of your hearing sensitivity rolls off as the years accumulate, undeniably and measurably so. As a result in crowded and noisy environments you will find it harder to distinguish what someone is saying to you because those high frequencies are used, if available, by the brain to better localise sound and to allow focus on a particular source. But all you will know is that you find it harder to understand what someone is saying. It will still seem like it sounds the same as it always did.

I, from personal experience, agree.

I think it is a good enough reason I'm utterly hopeless at decoding any language, not more than a word here and there, other than English.

For audio, I don't EQ to adjust anything for my hearing (or lack of).

My preferred EQ is for the direct/peak sound level (curve) at the listening position to match the peak curve of the source.

Recent simple example here.

I'm happy with the sound, nobody else complains or has any usablel suggestions, so it must be a reasonably correct method.

If I can't hear it, I can't hear it. Period. Nor can I even imagine what I'm not hearing, for the most part.
 

dasdoing

Major Contributor
Joined
May 20, 2020
Messages
4,209
Likes
2,674
Location
Salvador-Bahia-Brasil
Interesting. Now you've gotten me to thinking. If you read the fuller piece (posted after your comment) you'll see my entire argument. But could it be that one can have a two-track hearing process? Because, remember, the rest of the world is not loudness compensated.

If you live in front of your audio system (actually, I kind of do live in front of one of mine!) and don't experience other audio environments, then your brain's signal processor will probably adjust. Good point.

I'm just not sure that it's an adjustment that I want.

But perhaps one's personal signal processor can switch between modes.

I will focus on "artificial sounding"

no attenuation will ever sound natural, because how would you do it in the real world? let's say you are hearing an orchestra. how could you attenuate it? you can use ear plugs. will your volume attenuation sound like this? no. you could have the orchestra playing the piece softer. besides destroying the tonality of the original piece, can you reproduce this electronicly? again no.
so what would be the best option (*)? wouldn't it be attenuating while preserving the overall balance?

it sure sounds strange in the beginning...you are used to "breaking the balance" while attenuating for decades

(*) or let's say second best option, because the best will always be playing back at the same volume the engenier balanced the music to
 

Rock Rabbit

Active Member
Joined
Feb 24, 2019
Messages
230
Likes
174
is there a natural way of hearing an attenuated recording?
let's take a recording of an orchestra with a totaly flat mic. it will sound natural if you reproduce it at the exact same loudness level the mic "heard". if you change the level it can't sound natural anymore.
now, since you are an engineer. what if the mastering reference level for some reason was 60dB? the level would be lower than what was "heard" by the mic. would or wouldn't you equalize the balance in the master to compensate for that?
Attenuated recording? A pianissimo is at some 60 dB @ 1 mt. At what distance was the recording mic in an orchestra (?). Suppose you adjust level at 80-84 dBSPL, and then comes a fortissimo at over 100 dB ...but you added 20 dB to hear the "natural" sound! now you are hearing over the max sound of a piano at an unhealthy pressure level. For a more realistic example
https://academic.oup.com/view-large/3169083
In the case of multiple microphones recording...what is the "natural" loudness of the recording? Is this "average" loudness similar to perceived loudness by the audience? And then when the recording is reproduced by loudspeakers in a typical room, room modes can affect +/- 10 dB or more at low frequency spoling everything.
The idea of listening at same natural level as the real instruments is not practical...many wind and percussion instruments can go easily over 120 dBSPL, a Timpani 130 dB.
 

playmusic

Member
Joined
Dec 14, 2020
Messages
10
Likes
6
The idea of listening at same natural level as the real instruments is not practical...many wind and percussion instruments can go easily over 120 dBSPL, a Timpani 130 dB.
With audio it takes some effort to formulate in an accurate way. Here is my attempt.

Concerning health issues from sound, the usual measure is A-weighted dB SPL RMS.
In what follows, dB is an abbreviation for this, unless otherwise noted.

Musicians of a symphony orchestra are exposed at their ears to around 90 dB during a concert.
Exposure can be higher for a few instruments, e.g. trombones (95 dB) at the ear of musicians.

For short term (i.e. seconds), levels of up to 120 dB can be reached (e.g. trombones, drums) at the ear of musicians.

Exposure of the audience is (obviously) lower than for musicians and around 80 dB.
Peaks are around 100 dB SPL.
So it is practical (both from a health and a technical perspective) to hear music as loud as classical music concert goers.

I've been arguing for years that applying loudness compensation EQ is a fundamentally misconceived idea.
@Stephen Dawson: You might want to have a look at post 128 by @dasdoing in this thread.
In this case, the attenuated compensated music sounded to me better (in the sense of more similar to the original at high dB SPL) than the attenuated uncompensated version.
 

Aerith Gainsborough

Addicted to Fun and Learning
Joined
May 4, 2020
Messages
853
Likes
1,280
The idea of listening at same natural level as the real instruments is not practical...many wind and percussion instruments can go easily over 120 dBSPL, a Timpani 130 dB.
Listening to the instruments at their original value for the audience in a concert hall is most certainly practical.

If you do, your loudness compensation will do: nothing. You are listening at the calibrated level of ~85dB and don't need it. The EQ will be flat.

Naturally, we don't want to listen at 85dB all the time. I mean an orchestra is LOUD.
So we turn it down by 20dB. Easily done in the realm of electronics. Suddenly, bass notes that were audible before are now inaudible.

This is where loudness compensation kicks in and brings them back to potentially audible. Still quieter than @ 85dB though. So the bass sounds quieter but it is still there.

Is this kind of compensation needed? Technically no. Our ears and brains expect that the spectral balance gets out of whack when we turn it down. That's why many people love listening loud because then they get the feeling of "everything is as it should be".

Can it be enjoyable: absolutely, if you get the levels right, that is. If you don't... boomy bass awaits.
Nothing a quick grab at the shelving tone controls won't fix though.
 

dasdoing

Major Contributor
Joined
May 20, 2020
Messages
4,209
Likes
2,674
Location
Salvador-Bahia-Brasil
Attenuated recording? A pianissimo is at some 60 dB @ 1 mt. At what distance was the recording mic in an orchestra (?). Suppose you adjust level at 80-84 dBSPL, and then comes a fortissimo at over 100 dB ...but you added 20 dB to hear the "natural" sound! now you are hearing over the max sound of a piano at an unhealthy pressure level. For a more realistic example
https://academic.oup.com/view-large/3169083
In the case of multiple microphones recording...what is the "natural" loudness of the recording? Is this "average" loudness similar to perceived loudness by the audience? And then when the recording is reproduced by loudspeakers in a typical room, room modes can affect +/- 10 dB or more at low frequency spoling everything.
The idea of listening at same natural level as the real instruments is not practical...many wind and percussion instruments can go easily over 120 dBSPL, a Timpani 130 dB.

not sure where you are going here. a raw recording of an orchestra represents levels at the point the mic is. then it is mastered to represent "the perfect seat" in the audience. that means, the engenier will balance it at his reference level to sound as natural a possible. it doesn't mean that a pianissimo will be at 83dB, and there will sure be no instruments at 130dB
 

Rock Rabbit

Active Member
Joined
Feb 24, 2019
Messages
230
Likes
174
not sure where you are going here. a raw recording of an orchestra represents levels at the point the mic is. then it is mastered to represent "the perfect seat" in the audience. that means, the engenier will balance it at his reference level to sound as natural a possible. it doesn't mean that a pianissimo will be at 83dB, and there will sure be no instruments at 130dB
Simple: first you can't reproduce the same dynamic range in a room (a very different room) with different reverberation and modes. Second, the idea of loudness curve compensation is flawed by the room modes (easily +/- 10 dB peak variations). Furthermore this compensation curves are not linear, even at 80 phon +10 dB 100 Hz and +30 dB 20 Hz (healthy limit for ear), and are a simple statistic average and don't reflect your non linear real threshold (worst in a room with reflections). Finally the harmonics partials of any instrument are affected by the speaker room interactions plus the loudness compensation curve plus your real threshold ear curve, even with a perfect loudness variable compensation who can tell a good correlation with reality?
Spectrum2.gif

The speaker/room/loudness curve/ear can preserve the relative amplitude of harmonics of this instrument (viola)?
Is loudness curve eq + ear threshold flat?
 

dasdoing

Major Contributor
Joined
May 20, 2020
Messages
4,209
Likes
2,674
Location
Salvador-Bahia-Brasil
Simple: first you can't reproduce the same dynamic range in a room (a very different room) with different reverberation and modes. Second, the idea of loudness curve compensation is flawed by the room modes (easily +/- 10 dB peak variations). Furthermore this compensation curves are not linear, even at 80 phon +10 dB 100 Hz and +30 dB 20 Hz (healthy limit for ear), and are a simple statistic average and don't reflect your non linear real threshold (worst in a room with reflections). Finally the harmonics partials of any instrument are affected by the speaker room interactions plus the loudness compensation curve plus your real threshold ear curve, even with a perfect loudness variable compensation who can tell a good correlation with reality?
View attachment 102062
The speaker/room/loudness curve/ear can preserve the relative amplitude of harmonics of this instrument (viola)?
Is loudness curve eq + ear threshold flat?

the thing is, there is no perfect attenuation. nobody said the proposed method is perfect. what is beeing said is that it is much, much better then simply attenuating all frequencies equaly.
at this point it realy makes more sense to hear the result, then to argue over the theory.
if you have any method that does it even better, I will be happy to use it
 
Top Bottom