Should you use Fletcher-Munson loudness compensation?

[solderdude]

A basic loudness compensation circuit doesn't "know" the signal level or the acoustic loudness. It only knows the volume control setting.

Yes, this is the biggest issue for 'loudness' correction as was used on older stereo equipment.

The 'compensation' could be quite off when sources with lower or higher than expected levels was connected or when more or less efficient speakers were used. In those amps the volume control position determined the bass + sometimes treble boost.

It consisted mostly of a volume control with a center-tap halfway on the volume control where, using a switch, a network of capacitors and resistors were connected to it which created a tone control that varied with the volume control setting.
There were also more fancy version with 2 taps one at approx. 1/3 and the other at 2/3.

To do it properly one would indeed have to calibrate. The only variable then would be recordings. Classical recordings often have much lower average levels than some squashed popular music with a DR of 3 or lower.

 

[levimax]

The question is wrong. It should be should we go back to 60s understanding of music reproduction?

Are you saying the Fletcher-Munson curve is no longer valid Psychoacoustics? If so please link to the research.

 

[sarumbear]

Are you saying the Fletcher-Munson curve is no longer valid Psychoacoustics? If so please link to the research.

What you need to know is out in the open. Nobody agrees if such curves are correct.

Equal-loudness contour - Wikipedia

en.m.wikipedia.org

 

[DVDdoug]

It’s funny that there’s some decent consensus that loudness compensation is important but speakers aren’t built to an implied listening level.

Speakers are supposed to be linear and they are supposed to accurately reproduce the sound fed-into them. You wouldn't want the speakers to boost the bass during quiet passages or during fade-ins and fade-outs. But if you consistently listen at low levels you might prefer a speaker with boosted bass.

Should you compensate for this in your own listening? I’m aware in psychoacoustics the brain naturally compensates for many things

Your brain is doing the "wrong thing" when you listen at "artificially low" levels. If the music (or passage) is intentionally quiet, the "compensation" should be part of the performance or production.

what level is considered reference that you would compensate in relation to?

This paper is about mastering and it recommends monitoring at 79 or 85 dB so that's a range where mastering engineer should be is monitoring. But a lot of home listening will be at lower levels... We aren't always listening at "realistic" sound levels.

 

[abdo123]

I generated two filters that would transform a response that is flat at 80 Phon to flat at 60 Phon. Feel free to experiment.

 

[sarumbear]

I generated two filters that would transform a response that is flat at 80 Phon to flat at 60 Phon. Feel free to experiment.

View attachment 176274

Which curve you used?

 

[abdo123]

Which curve you used?

The red lines from the Wikipedia article you posted, feel free to suggest others. I'm trying to reduce my long term exposure to sound.

Will A/B with my headphones first, then check speakers.

 

[JeffGB]

I find the use of the fletcher-Munson compensation unnatural. If you listen to a band that has an average sound level of 95db and a few instruments are playing more softly than others, the compensation will be completely incorrect. If the one instrument is played at 75db and other instruments are played at 90db any compensation to boost the bass will distort the softer playing instrument if you are listening at 75db.

I fail to see how any use of the fletcher-Munson compensation can be accurate unless only 1 instrument/sound is playing and only then if the musician never changes the level of the notes they are playing.

 

[Chrispy]

I prefer to do so at my usual lower listening levels these days....matter of preference, tho.

 

[levimax]

What you need to know is out in the open. Nobody agrees if such curves are correct.

Equal-loudness contour - Wikipedia

en.m.wikipedia.org

I have seen this but all they are arguing about is that the exact details of the curve. I have never seen anything that contradicts the general concept.

 

[sarumbear]

The red lines from the Wikipedia article you posted, feel free to suggest others. I'm trying to reduce my long term exposure to sound.

Will A/B with my headphones first, then check speakers.

Fletcher Munson’s study was in 1933, later Robinson and Dadson in 1956 suggested new curves, which became the basis for an ISO 226 standard. I think if you want to use any curve it’s better to use the newer one.

Not that anyone agrees on those curves.

 

[abdo123]

Fletcher Munson’s study was in 1933, later Robinson and Dadson in 1956 suggested new curves, which became the basis for an ISO 226 standard. I think if you want to use any curve it’s better to use the newer one.

Not that anyone agrees on those curves.

The graph shows that the curves are from later revisions.

 

[sarumbear]

The graph shows that the curves are from later revisions.

Curves, curves, my kingdom for curves…

 

[levimax]

I think this discussion is getting lost in the weeds. There is not going to be any "perfectly correct" compensation as there are way too many variables. I do think that most people, if listening at very low levels, will prefer some bass (and possibly treble) boost. That is why tone controls are the best way to do this, add as much boost as subjectively sounds better to you and be done with it. Low level listening under any circumstances is not going sound as good as playback at louder levels.

 

[abdo123]

The experience is quite positive with A/B-ing with and without the Phon compensation. not absolutely perfect but i feel i can live with myself if i had to listen this way for the rest of my life.

@solderdude do we have any idea at what SPL the Harman research on headphones was conducted? I'm guessing the bass has a tiny bit of Phon compensation baked in.

 

[ZolaIII]

I really think it's type of reproducer and cone dependant. For the time when it whose originally done and the ISO revision it whose paper today it would be completely different (to the extent of material used to the extent of no harmonics at all with planars for example).

 

[solderdude]

@solderdude do we have any idea at what SPL the Harman research on headphones was conducted? I'm guessing the bass has a tiny bit of Phon compensation baked in.

I use Equal loudness contours for 70dB 'correction' in my plots and also my EQ as I rarely listen at 80 to 85dB.
I find it is better than the Harman response. The latter is more 'impressive' as bass is a bit 'separated yet boosted' which I find somewhat artificial and a tad too much. So a bit less than Harman and not as steep (so following loudness contours) somehow works better for me.
It is very much the same curve as the old Baxandall bass control set at +4dB at 20Hz just a little bit lower in frequency.

Sean Olive mentioned the levels that most tests are done at so it is known.
@GaryH might know the answer.

 

[abdo123]

I find the use of the fletcher-Munson compensation unnatural. If you listen to a band that has an average sound level of 95db and a few instruments are playing more softly than others, the compensation will be completely incorrect. If the one instrument is played at 75db and other instruments are played at 90db any compensation to boost the bass will distort the softer playing instrument if you are listening at 75db.

I fail to see how any use of the fletcher-Munson compensation can be accurate unless only 1 instrument/sound is playing and only then if the musician never changes the level of the notes they are playing.

loudness normalization can help with that.

 

[krabapple]

I’ve heard a lot of people state it sounds unnatural to them. Similar to how there’s an expectation of different sound in large vs small rooms, I could believe that equal loudness at lower levels sounds strange - in other words it might sound natural for music to be bass shy if it quiet.

But I don’t know either way which is why I wanted POVs.

There is the equal loudness curve(s) and there are methods for compensating. One is a fact and the others are methods for dealing with it.

I expect there is a big variation in implementation,

I didn't say all implementations are equally effective.

 

[krabapple]

What you need to know is out in the open. Nobody agrees if such curves are correct.

Equal-loudness contour - Wikipedia

en.m.wikipedia.org

'Everyone' agrees that bass (and high treble) balance goes off when levels are reduced.