What is the flattest target curve?

[Volutrik]

There seems to be a downward spiral that leads us to nowhere here. I've read and watched tons of reviews and articles, yet there seems to be no conclusion. Some people say that the Etymotic target is flat, some say that the Harman target is flat, some say that the IEF Neutral target is flat, some even say that the Harman target is purely a preference-based target curve (which isn't true, even Dr. Sean Olive stated this himself).

We all know different people hear things differently, but according to what I've learned, these differences aren't so drastic and unpredictable. They're a matter of some people having more or less of an already known frequency deviation, as of the example of high frequency hearing loss, thus leaving us to the conclusion that yes, it's possible to average human hearing.

Is there a real flat target curve or at least one that most closely represents real flatness? There are at least three target curves as cited above that claim to be flat. Which one is the real one? Would one be right for headphones but not right for in-ear monitors and that's why people misunderstand this subject?

 

[markanini]

The only target I can advise is markanini target: https://kurin.squig.link/?share=mar..._7Hz_Salnotes_Zero,Oratory_Moondrop_Starfield

 

[fieldcar]

just read this:

Graphs 101: How to Read Headphone Measurements – In-Ear Fidelity

A guide to reading frequency response graphs of headphones and IEMs.

crinacle.com

 

[GaryH]

There seems to be a downward spiral that leads us to nowhere here. I've read and watched tons of reviews and articles, yet there seems to be no conclusion. Some people say that the Etymotic target is flat, some say that the Harman target is flat, some say that the IEF Neutral target is flat, some even say that the Harman target is purely a preference-based target curve (which isn't true, even Dr. Sean Olive stated this himself).

We all know different people hear things differently, but according to what I've learned, these differences aren't so drastic and unpredictable. They're a matter of some people having more or less of an already known frequency deviation, as of the example of high frequency hearing loss, thus leaving us to the conclusion that yes, it's possible to average human hearing.

Is there a real flat target curve or at least one that most closely represents real flatness? There are at least three target curves as cited above that claim to be flat. Which one is the real one? Would one be right for headphones but not right for in-ear monitors and that's why people misunderstand this subject?

The only one of them that has empircal backing in the form of scientific, controlled, double-blind listening tests is the Harman target. And those blind tests also found that frequency responses with flat (raw response) bass and/or lacking in ear gain (as the Etymotic and IEF target do) were not well rated by listeners:

 

[Soria Moria]

Just go for a smooth frequency response closely aligned to the Harman target and then EQ it to be as neutral sounding to you as possible. Harman target sounds very neutral to my ears so I never deviate strongly.

 

[digitalfrost]

Is there a real flat target curve or at least one that most closely represents real flatness? There are at least three target curves as cited above that claim to be flat. Which one is the real one? Would one be right for headphones but not right for in-ear monitors and that's why people misunderstand this subject?

There is no perfection. Just start with something and see how that works for you. Can be Harman can be something else. In the end you need to tune to taste. We're all living the circle of confusion.

Target curves for headphones and in-ears are different anyway.

 

[markanini]

BTW Harman IE target was not as thoroughly tested as the over-ear version, but the premise is good. Where we are at now is numerous internal targets. That's why for IEMs I personally use a mix targets, Oratory1990 for lows, Harman 2019 for mids, and Moondrops for highs.

 

[GaryH]

BTW Harman IE target was not as thoroughly tested as the over-ear version

Yet still more thoroughly tested than any other proposed IE target, most of which have zero confirmed scientifically valid blind-test backing whatsoever, and as shown above Harman's tests suggest they would not do well once proper such controls are implemented.

 

[markanini]

But still more thoroughly tested than any other proposed IE target, most of which have zero scientifically valid blind-test backing whatsoever.

Assuming the statement in it's entirety is true, and it might be, the participants consisted of solely Harman employees and a fixed treble EQ was used, as stated by Dr. Sean Olive. Seeing past such limitations may be impractical in the real world, even irrational considering the possible big differences in possible outcome. At the very least, if you accept the limitations of Harmans methodology for the IE target, that leaves the door open to consider less well documented IE target efforts, by precedent, on some limited basis, such as I've done in lieu of greater public testing efforts that combine all of quality, scope and documentation.

 

[Hatto]

Only flat target is flat, any others are, well, NOT flat.

 

[DVDdoug]

Only flat target is flat, any others are, well, NOT flat.

That's what I was going to say... The flattest curve is flat. Most electronics is flat and that's what we want unless we are trying to make a correction with EQ or unless we like bass boost, etc.

It's hard to get flat from speakers & headphones, and a flat speaker isn't flat in a real-world room. And for whatever reasons, flat headphones don't "sound right" (to most listeners).

The preference curves are based on average preferences or what is "most commonly preferred".... Maybe nobody has the exact-same preference as you.

 

[markanini]

That's what I was going to say... The flattest curve is flat. Most electronics is flat and that's what we want unless we are trying to make a correction with EQ or unless we like bass boost, etc.

It's hard to get flat from speakers & headphones, and a flat speaker isn't flat in a real-world room. And for whatever reasons, flat headphones don't "sound right".

The preference curves are based on average preferences or what is "most commonly preferred".... Maybe nobody has the exact-same preference as you.

Flat is a great target for DACs, ADC, amps.

 

[GaryH]

the participants consisted of solely Harman employees

So what? It was double-blinded.

that leaves the door open to consider less well documented IE target efforts

It also leaves the door open to innumerable sighted cognitive biases and pseudoscience.

 

[markanini]

It also leaves the door open to innumerable sighted cognitive biases and pseudoscience.

You may use your judgment and make a personal decision. You propose extrapolating the the outcomes of one well documented, but limited study. I chose to consider the efforts of Harman and some lesser documented efforts on a more equal level, on a conditional basis. This means I dismiss any efforts lacking the necessary resources to replicate the quality of Harmans double-blind testing, which frankly isn't amazing for the their IE target due to the fixed treble, but nevertheless the highest documented effort for IEMs. This excludes reviewer targets, because they are created by single listeners. It excludes theoretical targets like sloped DF variations, because the impetus of Harmans work was to do with theoretical targets not matching listener preferences. It excludes some manufacturer targets like Etymotic because while moderately better documented there's no mention of double blind testing, fair to assume it didn't occur.

My personal target, a composite of existing targets, happens to be close to Harmans 2016 IE target. Some info about what happened in between that and the 2017 target would be interesting to learn about. And any more robust data to falsify you or me, would be welcome, but I don't hear of any such efforts coming soon.

 

[isostasy]

@Volutrik This subject has been brought up quite a lot already, there's been some very in-depth discussion on the 5128 thread which won't directly answer your question but will definitely make it apparent why we can't (and maybe shouldn't) agree on a single curve.

To answer directly some of your queries:

1. IEF Neutral is not flat, it's just crinacle's personal perception of neutral. It's a preference curve and you may or may not agree with it (I think it's lacking 2-3kHz). It also has no bass rise or shelf at all which most people would agree is required to some extent to perceive neutrality.
2. Etymotic is definitely not flat; following on from above, it has way too little bass. However many people, myself included, love the latest 'XR' versions of Etymotics which have a bass boost which goes most of the way of counter-acting this.

It depends why you're asking this question. Academic interest, shopping, etc. I think it's good there's more effort and interest put into this but for music enjoyment most of us overthink it.

 

[JJFUSA]

You may use your judgment and make a personal decision. You propose extrapolating the the outcomes of one well documented, but limited study. I chose to consider the efforts of Harman and some lesser documented efforts on a more equal level, on a conditional basis. This means I dismiss any efforts lacking the necessary resources to replicate the quality of Harmans double-blind testing, which frankly isn't amazing for the their IE target due to the fixed treble, but nevertheless the highest documented effort for IEMs. This excludes reviewer targets, because they are created by single listeners. It excludes theoretical targets like sloped DF variations, because the impetus of Harmans work was to do with theoretical targets not matching listener preferences. It excludes some manufacturer targets like Etymotic because while moderately better documented there's no mention of double blind testing, fair to assume it didn't occur.

My personal target, a composite of existing targets, happens to be close to Harmans 2016 IE target. Some info about what happened in between that and the 2017 target would be interesting to learn about. And any more robust data to falsify you or me, would be welcome, but I don't hear of any such efforts coming soon.
View attachment 277587

I believe, but I could be wrong indeed, that more important is to look how smooth the FR is rather than precisely follow any of the targets as all of them are quite similar in shapes.

Bass could be adjusted to the particular taste from neutral to elevated by 5 - 10 dB, 3 Khz range is naturally elevated by 10 - 15 dB, then slow treble decline, that's why cheap IEMs like 7Hz Zero and Hexa sound so good I guess.

 

[markanini]

I believe, but I could be wrong indeed, that more important is to look how smooth the FR is rather than precisely follow any of the targets as all of them are quite similar in shapes.

Bass could be adjusted to the particular taste from neutral to elevated by 5 - 10 dB, 3 Khz range is naturally elevated by 10 - 15 dB, then slow treble decline, that's why cheap IEMs like 7Hz Zero and Hexa sound so good I guess.
View attachment 277591

Everything looks smoother on Resolves rig, he uses the GRAS HR-coupler.

 

[dasdoing]

there have only been preference studies, so we don't know.
also we are translating a signal of 60 degrees into 180 degrees. it can't even be neutral in all pan positions even if we find (one) neutrality curve in an anachoic chamber.
at the end, listening to stereo on headphones is just something that can't work. binaural music should be the standard for headphones.

 

[digitalfrost]

The only target I can advise is markanini target: https://kurin.squig.link/?share=mar..._7Hz_Salnotes_Zero,Oratory_Moondrop_Starfield

View attachment 277397

Can you provide a CSV that I can use with AutoEQ?

 

[markanini]

Can you provide a CSV that I can use with AutoEQ?