3 Comments

I should have expected this. The original paper that eventually resulted in the Harman curve was published ten years ago. Like practically every paper on audio that takes a strong stand, it was met with a mixture of praise, scorn, and indifference. Now others are responding by suggesting their own alternatives to the Harman curve—and headphone enthusiasts are left scratching their heads, wondering which, if any, of these will agree fully with their own ears.

I’m aware of five different published modern, basic headphone target curves, although some of them have variants, and I know manufacturers all have target curves of some sort, even if they’re not published. (If you like, you could throw in the free-field and diffuse-field target curves, too, but they’re considered obsolete for the most part.) I hope the originators of these five curves will pardon any oversimplifications in my following brief synopses. Let’s begin with a chart I’ve concocted that compares these curves; the versions of the curves here are all intended for earphone voicing. It’s a relatively low-resolution chart that I created by hand, based on charts I’ve seen of the different curves.

Chart

First and foremost is the Harman curve, the product of a research project by Harman International that resulted in 19 papers and has been adopted in headphones from all the Harman brands (AKG, JBL, Harman/Kardon, and Mark Levinson). It was originally based on the sound of a good loudspeaker in a good-sounding room, and modified through extensive user-preference research. Many other manufacturers—including 64 Audio, Dan Clark Audio, PSB, and Puro—have spoken either of the Harman curve’s influence on their work, or of the similarity of the results of their own research to the Harman recommendations. A few audio publications use the Harman curve as a standard to judge headphones, under the notion that headphones and earphones that stick close to it are good, and models that stray significantly from it are not recommendable.

The Preferred Listening Response Curve from Knowles, a major manufacturer of miniature speakers (i.e., headphone drivers) and microphones, is basically an update of the Harman curve. The Harman curve was produced using headphone measurement gear designed to the original IEC 60268-7 standard, which delivered valid and consistent results up to only 8kHz. But five years ago, headphone-measurement gear manufacturers introduced “high resolution” ear simulators that produce consistent measurements up to 20kHz, or even higher. Knowles has used this equipment to conduct additional research that modifies the Harman curve with elevated treble levels above 3.5kHz. (You can download a white paper on the Knowles curve here.)

Ear Simulator

The Crinacle IEM Neutral Target is the creation of audio reviewer and consultant Crinacle. It is the product of his experience in reviewing and his perhaps-unmatched database of earphone measurements. As he describes it, the IEM Neutral Target is what he considers to deliver a “perceptibly neutral” response. I recently reviewed an earphone design that Crinacle voiced, and you can read my comments about the sound in that review.

Next up is the preference curve used by Soundguys, an audio-review site. I gather from reading the site that it represents an adjusted average of responses of headphones and earphones liked by the staff, and adapted to their specific measurement gear. As best I can tell (I don’t know how the compensations they made affect their curve), it’s somewhat similar to the Harman curve, but with less emphasis in the treble and bass (or more in the midrange, depending on how you want to look at it).

Last—because it’s the most recent to emerge—is Sonarworks SoundID SR, the creation of a company that makes calibration technology for headphones and studio monitor speakers. (I reviewed the Sonarworks headphone calibration software four years ago.) The notion is similar to that of the Harman curve, which used as its initial target the sound of a good set of speakers in an acoustically good room of the type that might be encountered in a typical home. Sonarworks proposes instead the target of a calibrated set of speakers in a well-designed studio control room. It also proposes that, because headphone measurements are to some degree inaccurate and inconsistent, all measured headphone responses should be confirmed by a panel of five listeners (either the manufacturer’s own panel or Sonarworks’ panel), comparing them to a set of calibrated studio monitor speakers. It also suggests that manufacturers should feel free to deviate from this curve as they see fit, but that their variance from the curve should be documented.

Five people

By now, you may be wondering which of these curves I endorse. The answer—and this has always been official policy at SoundStage! Solo—is none of them. I once proposed, at the suggestion of some readers, to reference all of our headphone measurements to the Harman curve, but SoundStage! founder Doug Schneider shot that idea down, and I’m glad he did. “How can we say that every other manufacturer should conform to Harman’s curve, no matter how good their research is?” he asked me, and I couldn’t make the case. I do, though, include in every comparison chart a measurement of a similar headphone or earphone that comes close to the Harman curve.

I heard two comments at the recent 2022 CanJam SoCal headphone show that confirmed my thoughts on this matter. One was from Dan Clark of Dan Clark Audio, describing how he voiced his new Expanse high-end headphones—with, as he explained thoroughly, a little more upper bass and a little less treble than the Harman curve. He made a strong case for his strategy, but more important, the Harman curve provided a basis for our discussion. It’s something both of us understand well, that we respect even though we don’t use it as a substitute for our own judgment.

The other comment came from a fellow headphone reviewer, someone who doesn’t do measurements but who nonetheless possesses a solid technical understanding of audio. Of the Harman curve, he said, “I wouldn’t say it’s perfect, but it’s a reasonable reference.”

Man sitting

Whether or not any of these target curves square with your own perceptions and opinions about sound is something only you can judge. Personally, I’ve found I can enjoy quite a few headphones and earphones that stray from industry convention in certain ways. But even though not every Harman-curve headphone has ranked as one of my favorites, I haven’t disliked any of them, and I’m thankful to have it as a “reasonable reference” that I can use in my work.

. . . Brent Butterworth
This email address is being protected from spambots. You need JavaScript enabled to view it.

Say something here...
Log in with ( Sign Up ? )
or post as a guest
People in conversation:
Loading comment... The comment will be refreshed after 00:00.
  • This commment is unpublished.
    Dustin · 1 months ago
    Very informative article, Brent,  You write a lot like that.  Not sure I agree with you on Doug's direction, though.  I feel like the Harman research is more complete and robust than anything audio reviewers have ever traditionally relied upon, so I think it would be as good as anything to form a reference out of.  For the other curves, I never knew about them until i read this article.  Oratory seems to have some very useful comments around this.  Let's learn more about them.  Cool, thanks.
  • This commment is unpublished.
    SEAN OLIVE · 1 months ago
    It's noteworthy that the headphone voiced by Crinacle you recently reviewed and liked does not meet the Crinacle IEM Neutral Target shown here.  It is not flat bass shelf that is closer to the Harman IEM Target. 

    Oratory already pointed out that the SoundGuys Target is based on measurements made on a B&K 5128 which uses a different ear simulator.  We have a B&K 5128 andI have measured headphones  on the B&K 5128 and  compared them to measurements made on the GRAS45CA-10 (with the HB 5001 Anthropometric pinna) and Todd Welti's original modified pinna which was used to develop the Harman Target.  The differences above 2 kHz are not insignificant and there are difference below 100 Hz related to leakage (for headphones) and acoustic impedance (for earphones).. What this means is you cannot compare target curves on the same graph made on different test fixtures.

    Cheers
    Sean Olive

    • This commment is unpublished.
      Doug Schneider · 1 months ago
      Interesting comments Sean. When I see the variances in the headphone measurements, because of the test setup, I always ask, "Are we CLOSE to a test setup that exactly mimics the human head and can be considered THE STANDARD?"
      • This commment is unpublished.
        Sean · 1 months ago
        That is something we are actively  researching. Since everyone's head, pinna and ear canals are slightly different the best we can do is aim for a test fixture that representd  a statistical average of what is measured on a human. 

         Good headphone design can  accommodate and/or compensate for some of these differences We see active headphones now that measure the response on listeners ears  and compensate for hese variations. 

        When we did our research on preferred target curves we used open back  replicator  headphones that minimized these acoustic interactions between listeners and we  controlled leakage effects so that listeners were easentually delivered the similar sound. This control exploans in part why we got reliable subjective data ehich we were able to correlate and predict with objective measurements. We also used a test fixture with a  proprietary pinna that gave a more accurate representation of human leakage. So tight controls on both subjective and objective measurements. 

        So if companies can design headphones that account for these variations  they have the best chance of satisfying the most listeners. Using a test fixture that can reliably measure what on average is delivered to listeners is an essential aspect of the process 
  • This commment is unpublished.
    oratory1990 · 1 months ago
    An important note is that the target curve used by soundguys applies to the B&K Type 5128, which represents an ITU P.57 Type 4.3 ear simulator (the most accurate, and most recent), whereas the other targets apply to measurement rigs representing Type 2 ear simulators.

    Meaning that the soundguys target can not directly be compared against the others, as the measurement apparatus that it is used on delivers different results, most importantly (in my opinion) at low frequencies, where it more accurately models the acoustic impedance of the human ear - more accurately than the IEC60318-4 coupler used for the P.57 Type 2 ear simulators.
    • This commment is unpublished.
      oratory1990 · 1 months ago
      The intent behind the soundguys target - as far as I know - is to have a version of the Harman Target but for a Type 4.3 ear simulator.  
      Harman's research was conducted on Type 3.3 ear simulators (the results of which can mostly be used with Type 2 ear simulators as well)

Latest Comments

Brent Butterworth 2 days ago Focal Utopia Headphones
@Bill SizerIt's not out of the question. But unfortunately, DCA usually arranges for about five reviews ...
Bill Sizer 6 days ago Focal Utopia Headphones
Nice, thorough review...much appreciated! Speaking of top-tier headphones, any chance you guys will be able ...
@a1982They're supposed to be sending me one ... I'll check on it.
Hi can you measure the 2022 utopia? no one had ever published details like csd ...
@JimmyYeah, I created one for Wirecutter that's an average of the noise canceling between 100 ...
Have you thought of a rating system for noise cancellation. Useful like your other scale. ...
@Brent ButterworthFair. Thank you for your time. Appreciative of advice.
Very informative, easy to read article. It's gotten hard to find reviews of any kind ...
@JonathanYou should read those reviews on this site, read some other reviews, and decide. I ...
@Brent ButterworthThose 4. What would you pick for you? I want sound most, but good cancelling. ...