Pink Noise sounds boomy with Harman tuned headphones.

[thewas]

Fwiw, I thought Dr. Olive had some interesting thoughts on these subjects in the following live stream...

Could you maybe just shortly title a few of them as it is a 2,5 hour video?

 

[solderdude]

No particular reason. It would just be interesting to see how well actual in-room measurements might compare with those derived from the anechoic measurements.

Averaged over how many different speakers and those different speakers in different rooms/listening positions.

I sure can't predict how speakers will sound in a room, or even my own room, but can pick ones that most likely will not be a great speaker to start out with.
That's what it is good for.
For headphones (assuming good seal) I can make a better prediction.

 

[ZolaIII]

The equal loudness normalization researches as there whose plenty of them and where methodically better (more consistent and better documented and without obvious flaws) came to in the low frequency
region all the newer data deviate
systematically from the equal
loudness contour by Robinson and
Dadson [ISO 226]. These differences are
nearly as high as 15 dB.

This researches are inconclusive and based still on small number of test examples (even taken cumulatively it's still a to small pattern) but still represents a much better and more representative data than Harman hedaphone research. Anyone trying to say how such is final and done is silly and childish (or selling a snake oil if you wish). There is no valid preference curve and until we are able to prove, correlate, and explain such big differences there won't be any. Of course that will need a lot more resserch and expanding beyond current research limits (in other scientific research spheres and disciplines in order to explain them before all).
Best regards and have a nice time.

 

[dasdoing]

the best indication that it is bass-heavy is listening to deep male voices. they just don't sound natural. raw cello recordings are similar

 

[ADU]

Could you maybe just shortly title a few of them as it is a 2,5 hour video?

Some of the more interesting questions and comments re preference, measurements, etc. begin at around 8:00, thewas. So that's probably where you could start. If you need a little more context though, then you could back it up a bit more, to around 6:20, or just to the beginning of the video.

I don't know what the skill/information level is of some folks here, so I've pre-cued the video below to 6:20, to include some of Resolve's introductory comments about the Harman research, which eventually lead to the questions re preference testing, and the circle of confusion at about the 8:00 timeframe. Hopefully this will help some.

 

[fpitas]

Keep in mind...Harman wanted a typical curve to design to, one that keeps the majority of buyers happy. Obviously some buyers want extremes (think: Beats), or sometimes less bass. They can't please everyone simultaneously.

 

[ADU]

Averaged over how many different speakers and those different speakers in different rooms/listening positions.

Enough to get a good and fairly representative sampling of the in-ear response of some of the better speaker designs and manufacturers.

I sure can't predict how speakers will sound in a room, or even my own room, but can pick ones that most likely will not be a great speaker to start out with.
That's what it is good for.

When you say "that's what it is good for", I assume you're referring to Klippel. And would agree that is one of the things it's good for.

For headphones (assuming good seal) I can make a better prediction.

And what would this prediction be based on?

 

[solderdude]

But what is 'enough', how many different speakers, different how, how many different rooms and how should they differ ?
Who is going to fund this ?
What measurement and reproduction gear is going to be used ?
Who (with what experience and background) is going to do this ?
What rooms will be used and how is this documented ?
What speakers should be used and why ?
What listening level ?
What specific music choices ? Only the ones used in Harman testing or other recordings as well.
Who are going to do the listening tests ?
What would be the speaker/listener distance, speaker wall(s) distance and angle, ceiling height ?
Nearfield or at a few meter and how does this correlate ?
Carpet or wooden (or vinyl covered) floors or floors with some carpets ?

When the speaker reference business has been sorted and validated a similar project for headphones based on the new found 'reference' ?


What do you expect to find ? Enough info to question Harman (would have to include preference in the study as well ?) or Klippel calculation method ?

 

[Sokel]

Preference can never be law.The word tells everything we need to know.
I like harman curve with classic (90% of what I listen to) but that depends.Old recordings favour it,new ones of complex plays definitely needs a flatter response to my ears.
Again,preference.

 

[ADU]

But what is 'enough', how many different speakers, different how, how many different rooms and how should they differ ?
Who is going to fund this ?
What measurement gear is going to be used ?
Who (with what experience and background) is going to do this ?
What rooms will be used and how is this documented ?
What speakers should be used and why ?
What would be the speaker/listener distance, speaker wall(s) distance and angle ?
Ceiling height ?
Carpet or wooden (or vinyl covered) floors or floors with some carpets ?
What do you expect to find ? Enough info to question Harman (would have to include preference in the study as well ?) or Klippel calculation method ?

I believe that modeling of this kind is included in Harman's loudspeaker spinorama measurement methodology, and also in the design of professional spaces used for sound mastering. So those might be a few places to look for some of the other answers above.

I've also previously discussed the kinds of speakers that I think would be good candidates for something like this in other topics here. And they would fit the same basic criteria that others (such as ASR, and Doctors Toole and Olive of Harman) use for well-designed and well-measuring speakers. Namely, a flat direct/on-axis response, absence of resonances, smooth and fairly linear directivity and off-axis/sound power response, etc..

I might use slightly different groups of speakers, depending on the type and capabilities of the headphones I'm trying to tune or EQ though. For closed headphones, I'd probably use speakers that extend more into the sub-bass, or use a sub-woofer, for example. For open dynamic headphones, with less ability to extend into the lower frequencies, I might use a group of speakers with more average extension, and no sub. Or a mixture of speakers with good and less good extension.

 

[ADU]

Still interested in hearing some of your thoughts on this, if you feel like responding, solderdude...

For headphones (assuming good seal) I can make a better prediction.

And what would this prediction be based on?

 

[ZolaIII]

@ADU let's simplify this. When bass level (hottest part of signal) overlaps with mids and blure signal from there to let's say 5 KHz which is the area that human hearing is most sensitive and where 65% of all information is contained in known recordings it simply becomes bad for me (and I assume pretty much everyone else). However it's shown especially in newer equal loudness normalization studies how values under 500 Hz will vary a lot (up to 15 dB) from ISO 226. What needs to be done with serious and significant statistical pattern large scientific research is; why, how and how much depending of various most significant factors (demographic, age, hearing state...). As that is not believable to happen anytime soon because of cost of such (along with time needed for explaining and squishing differences and anomalies) and probably until it's obligated by low only thing what remains to us is individual preference and no it's not Harman or any other so called "preference curve". I see Sennheiser IE 80 S IEM's as a novel try in regards of this and certainly a thing that should be taken in account along with desired/programmed loudness level's when making digital EQ's towards our individual preference.

 

[ADU]

I will look at both your posts above again, and see if there's anything I can contribute, ZolaIII.

I understand that IEMs have a few more challenges than over-ear headpones, in terms of tuning. And that there may be some more advantages in doing in-ear measurements on the actual wearer in this situation, versus using a more generic measurement rig that is based on an average anatomy. And have also suggested some other approaches that can people try in their own homes to possibly accomplish something like this a bit more easily, using speakers that are EQ-ed to a flat response at the listening position. And then measuring their in-ear response, and combining that with an appropriate room curve, based on their preference, or the average in-room responses of well-measuring speakers.

The only application I personally have for equal loudness, or Fletcher Munson at the moment is to compensate for different loudness/listening levels. I think you're trying to make some other points though with that, which are related more broadly to individual preferences. So I'll re-read, and try to understand what you've said on that a little better.

 

[ZolaIII]

@ADU if our preference does differ so much under 500 Hz and apparently it does (I used more recent studies of equal loudness normalization just to show it) and it does depend of listening level also (which will vary a lot depending on pretty much everything) until it's better scientifically studied we don't have much choice really then to fall back to individual preferences (which really isn't a end of the world as it influences mostly only that region). It needs to be a study on at least 1M of samples (with wide un influenced demographics and with squished (and explained) anomalies to 1% so that we can even take about some baseline preference suitable to n%.
Best regards.

 

[markanini]

Pink noise is kind of boomy in a room with a sub, or floorstanders with large bass drivers. Harman is not targeting the sound of bookshelf speakers.

 

[solderdude]

Still interested in hearing some of your thoughts on this, if you feel like responding, solderdude...

FR (tonal balance, even-ness, dips, peaks), distortion at different levels, seal (when actually measured).

As there is no 'room with shitloads of variables that can change the sound of speakers, it is much easier to predict tonality.

 

[ADU]

I understand that IEMs have a few more challenges than over-ear headpones, in terms of tuning. And that there may be some more advantages in doing in-ear measurements on the actual wearer in this situation, versus using a more generic measurement rig that is based on an average anatomy. And have also suggested some other approaches that can people try in their own homes to possibly accomplish something like this a bit more easily, using speakers that are EQ-ed to a flat response at the listening position. And then measuring their in-ear response, and combining that with an appropriate room curve, based on their preference, or the average in-room responses of well-measuring speakers.

Although I can see potential advantages in using a more personalized measurement approach with IEMs, I suppose an approach like the one described above would probably still work better with over-ear headphones. Because an in-ear mic would have to be inserted pretty far inside the ear canal near a person's eardrum to really capture enough useful frequency data from an IEM for comparison with the in-ear response of a loudspeaker. Which would involve greater potential risks and damage to a person's hearing.

I am not that knowledgeable on the subject of IEMs (that's really more Crinacle's turf), but I suppose if you wanted greater personalization in the tuning of an IEM, then the only other alternatives I can think of would be approaches that involve virtual modeling of the ear. Or doing the comparisons and tuning/EQ adjustments by ear. I think former Sennheiser engineer, Axel Grell, may be working on some new projects to possibly help facilitate something like this.

 

[ADU]

Harman is not targeting the sound of bookshelf speakers.

Agree... At least not with their current targets anyway.

 

[ADU]

FR (tonal balance, even-ness, dips, peaks), distortion at different levels, seal (when actually measured).

As there is no 'room with shitloads of variables that can change the sound of speakers, it is much easier to predict tonality.

And yet, this is still the standard that most of the current (and also some former) luminaries in the field of headphones seem to prefer and recommend.

And it is also the type of response that most headphone listeners prefer, if you believe and accept most of the Harman research on this subject (and Doc Olive's remarks about it in the livestream video posted above).

 

[solderdude]

About headphones (not going to comment about speakers):

There are several 'measurement fixture standards' and then there is the Harman target which was created on a non standard fixture based on earlier speaker research.
The latter is a target that about 60% of the tested listeners seem to prefer. 40% has a different opinion/taste.
I have no idea if this is also true for speakers. That has been researched enough by Harman and others.

Could such headphone research be repeated by others ? Sure. Who is going to pay for that ?

Will this give us a different target ? Yes, guaranteed it will be somewhat different when one works from a different premise.
I mean even the headphone Harman target has been revised. (not talking about the used test fixture correction b.t.w.)
When the research is done based on preference then you will see elevated bass again, for certain, but will likely differ from Harman's.

For instance the 'bump' Dan Clark prefers actually makes sense to me as being a deviation from Harman 'bass boost' that is.
As the bass boost type in the original Harman research was fixed in spectrum but could only be varied in amplitude this is a bit suspect. Who made that choice and why ?
Maybe (well not maybe...maybe) the slope and or frequency will differ from the one 'invented' by Harman research depending on the goal of the research.
My 'bass correction' that I apply, as my own target for headphones, is less steep (closer to that of DCA in shape) but also a few dB less. So while that type of bass boost is impressive to listen to, for me being in the percentage that feels it is too much and would like to see a different slope, a small improvement could be made.

The question then is... if that were done and a preference test would be repeated with a different slope and lift will the percentage of people with a preference go up a notch or go down a notch. And if people are asked whether they prefer Harman or 'the new' target which side will the scale point to and how much ?
Alas we will never know as I don't think much research, on such a large or larger scale with even more 'variables' than a fixed frequency/slope bass and treble correction being offered, is going to be expensive, may require more accurate test fixtures (which will still improve) and may not be compatible with 'older' standard test fixtures.
Nah... we are going to have to work with that what has been done already and use that as a guideline to start from and not gospel and the only absolute truth. But having a standard to use with would be nice. There are those (a.t.m.).

Then there is the measurement method and seal which can throw in some of dirt for closed and semi-open headphones.

Still... for headphones I can predict better how it will sound to me than I can make an accurate prediction how a measured loudspeaker will sound in my room to me.
Regardless what 'target' is used in the measurement and when looking at multiple measurements on other than specific 'standard' test fixtures.

In the meantime it is best to use something as a standard and IMO the Harman target is closer to 'reality' than most other 'targets'.
Well maybe except for the 'optimum hifi' curve (Oratory uses) but that's because that is closer to my own curve (for me).
The difference between optimum hifi and Harman is audible but not shockingly so. Subtle at best in the preference range as it were.