[sander99]

Why not? I exposed above my considerations: Flat on headphones sounds closer to my speakers than Harman target" corrected.

I could interpret the above sentence in various different ways. To make things more clear: Do you have some specific headphones in your mind (which ones?), and do you mean that they sound closer to your speakers without EQ-ing them towards the Harman curve (entirely possible)? And/or do you mean that they measure absolute flat (not relative to some target response curve)(not very likely I think)?

 

[hakunamakaka]

It's from this paper I think:

Segmentation of Listeners Based on Their Preferred Headphone Sound Quality Profiles​

From memory they end up with 64% out of 130 listeners being happy with the target(because it's the one they picked as best within X headphones?), and like 20% picked others but are grouped as basically wanting less bass because that's what the trend shows for the FR of the headphones they preferred. while the rest is said to want more bass(the smaller group but for similar reasons).
I wouldn't bet money on me and that summary being entirely right, but that's how I seem to remember it.

So harman curve was matched against Original headphone tuning or something specifically made for this test ? Either way this is nothing as it is a very small number to make any conclusions from the test. It doesn't even cover huge range of headphones. Taste in music and preferred genres can be very dependent on location and time. To have valuable findings you would need at least 100k or even million of user inputs. It is possible through web poll where users could compare their headphone against the sound matched towards harman curve. You would be surprised that some of the headphones would play in favor for their original tuning than the one matched towards harman curve

 

[castleofargh]

Why not? I exposed above my considerations: Flat on headphones sounds closer to my speakers than Harman target" corrected.
I don't see the point. So some random people "voted" that this "sounds better" and no we are taking it as gospel? People that for all purposes might have been used to listen to Spotify free, with iPods or Beats?

Those graphs you posted are clearly a sort of raw measurements based on an ear simulator rig. They are not supposed to be flat for something that feels flat to the ear. I asked that question to you because I was wondering if maybe you got the wrong idea about those graphs as it is not specifically written on them if or what type of measurement it is(although a big bump near 3kHz is usually a solid indication of RAW measurement on an ear simulator).

Maybe read this and see if it helps.
https://www.stereophile.com/content/innerfidelity-headphone-measurements-explained

Pay particular attention to this:

The green line is supposed to illustrate what calibrated speakers in a pretty close to ideal room would look like on that measurement rig. One of the hypothesis from the guys at Harman at the beginning was that listeners would prefer that sound from headphones(kind of like you seem to do), so they took a HD800 and EQed it to give(on the dummy head) the same FR as their flat speakers in their listening room.
That particular listening experiment turned out to sound too bright to listeners. But even then you can see that the black line, which is the initial Harman curve, isn't all that different from the sound of flat speakers. Because of course we wish for a sound that feels pretty neutral. but it doesn't look like a flat line. It isn't supposed to.

 

[someguyontheinternet]

So some random people "voted" that this "sounds better" and no we are taking it as gospel? People that for all purposes might have been used to listen to Spotify free, with iPods or Beats?

Harman used groups of people with different experiences and found that rankings were stable across experience groups. The experience of the listener did not seem to have any bearing on the order of the ranking, but did seem to have an effect on how far apart the rankings would be.
For example the random person would give scores between 4 and 8 while the trained listeners would give scores between 2 and 9. However the order of the ranking was stable between experience groups (also nationality and age groups).

 

[gregorio]

[1] Also HTRF is the same for over-ear headphones and speakers, because it's the same ear (not IEM's). [2] As for bouncing on walls and body... that is not considered while mastering, don't they use they use near-field monitors?

1. The ear and HRTF is the same but the sound reaching the ears is different. Much of what's reaching the ears when listening to speakers is reflections ("bouncing on walls" and other surfaces) and those reflections have relatively little high freqs, unless the walls, ceilings and floor of your listening environment are made of glass or ceramic tiles. In addition, there's the direction of the sound with the speakers in front of you and the absorption and reflection from your skull and pinnae.
2. That's not correct. Firstly, mid-field monitors are almost always used while mastering, although some mastering engineers may also have near-fields for reference/checking. Secondly, even when using near-fields, you still get reflections, plus the effects of the body and the skull and pinnae.

[1] Makes no sense to me. Anyway, what do I know?
[2] I just tested the Harman target curves and, to me, they sound bad.

1. That's because you don't know or don't understand the science/reliable evidence.
2. There's three possible reasons for that:
A. You haven't applied the target curves correctly.
B. You fall outside the probability distribution for whom the Harman target applies. EG. You are not among the 64%.
C. You have little/no access to a reliable reference and value a type of sound signature that's actually relatively poor. Therefore "good" is incorrectly assumed (and subjectively perceived) as "bad" and vice versa. This is particularly common in the audiophile world, because of all the misleading marketing audiophiles are subjected to, which affects their sound quality/value judgements and perception.

[1] My theory is that people's brains get used to certain equipment response curve and then, when presented with something better, they feel that they need to bring it down to their brain comfort level, where they are familiar with. [2] I guess for me, that "average" of opinions doesn't cut it. [3] It would be interesting to know what was their home equipment and music sources.

1. Sure and there's science/reliable evidence to support that theory. In fact, it's quite common that audiophiles exposed to a much more accurate system (say a commercial studio for example) will criticise it, relative to their own, poorer system.
2. Maybe, as per "B" above. Or maybe you are subject to "C" above, and that "average of opinions" would "cut it" if you re-educated/acclimatised to a better reference.
3. It varied and in some cases would not be relevant anyway, because some of the subjects were sound/music engineers accustomed to very high quality listening environments.

Why not? I exposed above my considerations: Flat on headphones sounds closer to my speakers than Harman target" corrected.

That's very possibly because the Harman target is designed to get closer to the freq response of the sound hitting your ear drums from an accurate system/listening environment, it's NOT designed to get closer to YOUR speakers!

[1] I don't see the point. So some random people "voted" that this "sounds better" and no we are taking it as gospel? [2] People that for all purposes might have been used to listen to Spotify free, with iPods or Beats?

1. Firstly, no one is taking it "as gospel". We're taking it as (by far) the most reliable evidence that applies to the majority of people (but not necessarily all). Secondly, you don't seem to realise you are contradicting yourself. Presumably because you don't understand what "random" means in the context of a scientific study or group of studies. Studies are commonly deliberately "randomised", so that different demographics are represented. EG. Members of the general public, professionally trained listeners such as engineers and musicians, younger and older people, males and females, etc. ...
2. This is the contradiction: You stated "random people" but then stated that the test subjects "might have been used to Spotify free, with iPods or beats", which isn't random people, it's a specific sub-group! As the test subjects were randomised (over various studies), some of them probably were "used to listen to Spotify free, with iPods or Beats" but certainly not all or even the majority. Others were used to listening to high quality systems and even professionally tuned systems/environments.

Curious how they come up with 64% of people preferring harman curve. It can be very dependant on the music genre that person listens.

No it can't, because that 64% was arrived at across a range of different genres.

Matching different set of headphones is very limited and in my view is only possible with the ones that sounds similar to begin with

You seem to be missing the basic fact that the Harman target curve is for the HP output, not the input!

To have valuable findings you would need at least 100k or even million of user inputs.

That's nonsense. One can have "valuable findings" from a sample size of just 10 or so test subjects. However, one would not necessarily have high confidence that those "findings" applied widely. As sample size increases so does the confidence level but this relationship is NOT linear, there are diminishing returns beyond a certain point. This is a very well studied area in science because huge amounts of money and countless lives are dependant on it: Market research for products, advertising and politics, plus drug/medical research for example. In most cases "valuable findings" with a high level of confidence can be obtained with sample sizes in the few hundred to a few thousand range and even in the most extreme cases, where lives are at stake throughout the globe (phase 3 vaccine and drug studies/trials for example) then roughly 1,000 - 20,000 is sufficient. So where do you get your 100k - 1 million from? Again, just another example of you simply making up whatever supports your agenda, regardless of whether it contradicts the demonstrated/established science!

@SoNic67 looking at the demographics of the Harman curve, you would seem to fall into the 21% sub-group of older (or female) subjects, who demonstrated a perception of significantly lower requirement of bass and therefore inversely, more high bass, mids and treble. In other words, the Harman curve for your demographic is very roughly about 5dB lower than the general curve.

G

 

[SoNic67]

The green line is supposed to illustrate what calibrated speakers in a pretty close to ideal room would look like on that measurement rig.

Harman used groups of people with different experiences and found that rankings were stable across experience groups.

See, that's what I don't get. IMO that line is not how speakers are "calibrated", but how they said their averaged group preferred to hear them. To me that's not only unbelievable, but when I applied those correction to my headphones, it sounded bad.

As all those people preferred to hear the voce clear and hated the cymbals/high-hat/tambourines, etc... That's probably formed by years of listening to mp3's (and similar lossy encoded) and that formed their brain into thinking that' show music should sound! Music is not only ears and transducers, is also brain.

Do you have some specific headphones in your mind (which ones?), and do you mean that they sound closer to your speakers without EQ-ing them towards the Harman curve (entirely possible)?

I have tried the curved on Grado SR 60i, AKG K701, and HIFIMAN Deva Pro.
Compared to what I hear trough my Klipsch RM-51M speakers.

I know that some people say that the horn-loaded AL tweeters on Klipsch sound "too harsh", but then they also probably never listened to a live rock band, to hear how those cymbals should really sound. Again, the same brain bias of what are you used to hear...

That's because you don't know or don't understand the science/reliable evidence.

Maybe. But to me this doesn't sound like science, it sounds more like a voting system. 130 people voted, and we are all supposed to conform to the result of that "voted" result. That's not really science, is more like sociology. It's more of how people brains got mangled by listening to decades of poorly compressed streamed music, FM radio and mp3's. Ah, and bad SiriusXM (my case, when I drive). Plus response in the cars are really screwed up - I have tested my car JBL/Harman system frequency response and it had a "bump up" at about 8kHz. People in US get probably used to that kind of response too.

I didn't see any "science" that can explain me those two bumps, besides statistics. The mixing in studio (near or mid field) is not that different from home or headphones!
If it was it true, it would imply that, at our hoses, or in our cans, the mids are attenuated and the treble are accentuated compared to studio and the curve reverses that. No way that's true!
PS: I typed above my equipment.

 

[someguyontheinternet]

As all those people preferred to hear the voce clear and hated the cymbals/high-hat/tambourines, etc... That's probably formed by years of listening to mp3's (and similar lossy encoded) and that formed their brain into thinking that' show music should sound! Music is not only ears and transducers, is also brain.

I don't think that this is reasonable to assume, because the test groups included audio-salespeople, audio gear reviewers, trained listeners and "normal" people. The experience range was quite varied so this MP3 assumption is not likely to hold.
The research shows that between these groups the ranking order of headphones was consistent. Every group put the same headphone at 1st place, 2nd place, etc down to last place.

 

[gregorio]

See, that's what I don't get. IMO that line is not how speakers are "calibrated", but how they said their averaged group preferred to hear them.

That's why you don't get it! You are talking about the speaker output while we (and Harman) are talking about the ears' input, two completely different things! The green line represents the ears' input when the speakers are calibrated flat. Castleofargh has already explained this to you. In other words, that green curve is typically what arrives at the ears and is what the brain perceives as flat.

[1] To me that's not only unbelievable, [2] but when I applied those correction to my headphones, it sounded bad.

1. The way you seem to understand it, is unbelievable to everyone else too! The problem isn't with the science but of your understanding (misunderstanding) of it.
2. And how did you apply it to your HPs? The amount of correction required would obviously depend on the freq response (output) of your HPs to start with.

Maybe. But to me this doesn't sound like science, it sounds more like a voting system.

But it is science, so if it doesn't sound like science to you, then clearly you don't know what science should sound like. Admittedly, psycho-acoustics often isn't as clear cut as some other scientific areas but it's still science!

123 people voted, and we are all supposed to conform to the result of that "voted" result. That's not really science, is more like sociology.

Firstly, 123 people didn't vote. The Harman curve is the result of multiple papers/studies. Secondly, we are NOT all supposed to conform to that result, only 64% are. Again, this has already been explained to you!

It's more of how people brains got mangled by listening to decades of poorly compressed streamed music, FM radio and mp3's. Ah, and bad SiriusXM.

Now you're just making up nonsense to support your misunderstanding. Unless of course you've got some reliable evidence to support your assertion? Again, some/many of those tested were professionally trained listeners, whose "brains got mangled" by working for many hours every day of the week with high quality systems/monitoring environments.

I didn't see any "science" that can explain me those two bumps.

Exactly! There is in fact old, very well established and uncontested science. So clearly, the issue is what YOU "didn't see", and NOT the science itself! For example, the first bump is due to the resonant frequency of the ear canal, this has been known and demonstrated/proven for roughly a century or so.

The mixing in studio (near or mid field) is not that different from home or headphones!

It depends on what you mean by "that", there certainly are significant differences. Have you ever spent much time in a high quality commercial studio?

If it was it would imply that, at our hoses, or in our cans, the mids are attenuated and the treble are accentuated compared to studio. No way that's true!

How does it imply that?

G

 

[SoNic67]

The green line represents the ears' input when the speakers are calibrated flat.

That is not ears input, that's just a voted preference. It means squat, for reasons that I described - people used to FM radio, streaming lossy files...

For example, the first bump is due to the resonant frequency of the ear canal, this has been known and demonstrated/proven for roughly a century or so.

The correction curves actually ACCENTUATE that frequency. If it was to correct any resonances, the curve would have to be the other way!
And there is no reason in the curve for the DROP after the 10 kHz.
That's the target that is "desired" by those EQ programs:

T

 

[hakunamakaka]

That's nonsense. One can have "valuable findings" from a sample size of just 10 or so test subjects. However, one would not necessarily have high confidence that those "findings" applied widely. As sample size increases so does the confidence level but this relationship is NOT linear, there are diminishing returns beyond a certain point. This is a very well studied area in science because huge amounts of money and countless lives are dependant on it: Market research for products, advertising and politics, plus drug/medical research for example. In most cases "valuable findings" with a high level of confidence can be obtained with sample sizes in the few hundred to a few thousand range and even in the most extreme cases, where lives are at stake throughout the globe (phase 3 vaccine and drug studies/trials for example) then roughly 1,000 - 20,000 is sufficient. So where do you get your 100k - 1 million from? Again, just another example of you simply making up whatever supports your agenda, regardless of whether it contradicts the demonstrated/established science!



G

Now you comparing against drug tests which has general use case, but human brain works differently than organs.

When peoples taste and preferences are involved you actually need way larger amount of inputs. Check how IMDB's movie scores fluctuates when they have small amount of voters. Plus you need to multiply it against all various types of headphones and music genres available, because conditions will change once you swap gear and genre. You need to be very specific with an outcome too.

The only route to find the most enjoyable sound is individually through trial & error. There isn't any test or machine available which would be able to tell you what fits you best

 

[SoNic67]

What I don't get is why are we accepting that albums mastered in studio have to be rendered on our systems with higher mids and lower treble.
To me, the excuses that our "home" equipment is not adequate (really, are we even going there?), or that our ear channels are worse than the ones of audio engineers that mastered that music (mine have "resonances", but theirs don't), are just not good enough.

Just because a bunch of random people decided that's how they like their music. And we call that a "standard".

 

[gregorio]

I know that some people say that the horn-loaded AL tweeters on Klipsch sound "too harsh",

Missed this previously. I actually heard Klipsch "Reference" speakers not that long ago and thought they were pretty good for home/consumer speakers, although I agree they're maybe a little too harsh. Horn tweeters make perfect sense for far-field monitors/speakers, not so much for home use though.

but then they also probably never listened to a live rock band, to hear how those cymbals should really sound.

How much time have you spent listening to a live rock band, both in a studio and on stage, before recording/amplification/mixing and after? How much do you know about "how those cymbals should really sound"? If you're going to attempt an "appeal to authority" then at least make sure you have more than those you're arguing with. However, as an "appeal to authority" is a fallacy anyway, it's obviously not acceptable as an argument to start with!

Again, the same brain bias of what are you used to hear...

And therefore, also the bias of what you are used to hearing, which according to you are speakers, that are somewhat harsh!

That is not ears input, that's just a voted preference.

No it's not! Again, the green line represents the in ear (dummy head) measurements of flat speakers in an acoustically treated room. The black line represents the preferred HP output, that closely correlates with the flat speakers in ear measurements. How many times?

What I don't get is why are we accepting that albums mastered in studio have to be rendered on our systems with higher mids and lower treble.

No one else gets that either and the reason is because no one apart from you is suggesting that! The Haman curve is for headphone (only) output in order to perceptually have the SAME bass, mids and treble as would be measured in ear in a studio (treated acoustic room). This is acheved by having slightly lower mids (not as you falsely state higher mids) and treble. You see that the black line is lower than the green line in the mids right?

To me, the excuses that our "home" equipment is not adequate (really, are we even going there?), or that our ear channels are worse than the ones of audio engineers that mastered that music (mine have "resonances", but theirs don't), are just not good enough.

What excuses, the one's you've just made up and are arguing with yourself about?

Just because a bunch of random people decided that's how they like their music. And we call that a "standard".

Yes, just because a large bunch of randomised people under controlled conditions reported their preference/subjective observations. And "yes" that's exactly how we define many audio standards, even some internationally agreed and legally binding ones!

G

 

[SoNic67]

If you're going to attempt an "appeal to authority" then at least make sure you have more than those you're arguing with.

LOL, appeal to authority. Let me try that for real then:
So. based on your arguments, the mixing engineers have no clue what they are doing, those treble need to be squashed by -10dB and the mids have to be elevated +10dB.
Gotcha.

And "yes" that's exactly how we define many audio standards, even some internationally agreed and legally binding ones!

Should we include Flat Earth Society in voting if Earth is round?

The Haman curve is for headphone (only) output in order to perceptually have the SAME bass, mids and treble as would be measured in ear in a studio (treated acoustic room).

They were not result of any measurements. It's just a "preference" curve.
And the studio engineers don't mix for treated rooms, they know their job.
Also... I thought this was about headphones? Hos the treatment of a room would affect them?

 

[gregorio]

LOL, appeal to authority. Let me try that for real then:

Nice try, again. But unfortunately you've got it badly wrong, you didn't listen to the fact that an appeal to authority is a fallacy and is therefore not acceptable here AND you didn't listen to the advice to make sure that you actually are the authority, because ...

So. based on your arguments, the mixing engineers have no clue what they are doing, those treble need to be squashed by -10dB and the mids have to be elevated +10dB.

Err, I am a mix engineer, that's how I've earned my living for nearly 30 years. I do know what I'm doing and I don't need to squash the treble by 10dB or boost the mids by 10dB because room reflections, my body, skull, pinnae and ear canals already do that. Again, you don't seem to understand even the basic concept of the Harman curve or anything about psycho-acoustics.

Should we include Flat Earth Society in voting if Earth is round?

Sure, if you believe that the earth being round is just an aural perception? The rest of us rational people do not believe that though, because of the science/reliable evidence!

They were not result of any measurements. It's just a "preference" curve.

Wrong, have you even read what the Harman curve is and what the green line represents?

And the studio engineers don't mix for treated rooms, they know their job.

Yes, I obviously do know my job, what do you know about my job?

Also... I thought this was about headphones? Hos the treatment of a room would affect them?

Huh, that's what the Harman headphone curve is based on. Commercial music/audio is created in treated rooms, that's what recording and mastering studios are, didn't you know that?

G

 

[castleofargh]

See, that's what I don't get. IMO that line is not how speakers are "calibrated", but how they said their averaged group preferred to hear them. To me that's not only unbelievable, but when I applied those correction to my headphones, it sounded bad.

I don't know what to say to this. How is your opinion relevant about what that graph actually is? Right below the graph in the article from Tyll(I linked and suggested you read), you can find:

Green dashed line is the ear drum response of a speaker that measures flat in the room. Black line is the subjectively derived preferred ear drum response for headphones.

As annotation to go with the graph.

And below he goes again:

The graph above shows the ear drum response as measured on a dummy head at the normal listening position between a pair of speakers. The green dashed line shows the ear drum response for a speaker that has been equalized flat at the listening position. The black line shows the adjustment away from flat while wearing headphones that most people chose as more pleasing.

Clearly the green line has nothing to do with group preferences, the black line does(although it was only an early version of the curve). The paper discussing that particular experiment to setup a listening test with various targets, DF,FF, the default FR of the headphone, and that flat speaker calibration being neither free of reverb nor overly reverberant(that would then measure as the green line when applied to the headphone) is called:

Listener preference for different headphone target response curves​

I can't find it in open access. But the test itself just concluded that people preferred the frequency response like that of the flat speakers over other older standards. And I don't know if it was in this paper or later, they seemed to find it too bright. better than the other options, but too bright. That's only because of that realization that the black line came to be a thing.

So far you've misinterpreted and probably badly tested a bunch of things, I hope that this can somehow point you in the right direction. Because regardless of enjoying the Harman curve or not, which is nothing more than personal taste, if you don't understand the data at all, how can you pass any sort of judgement about it?