Harman Curve Poll (Public)

Discussion in 'Blind Testing and Psychoacoustics' started by ChaChaRealSmooth, Nov 9, 2020.

?

What do YOU think of the Harman Curve

  1. Strongly Disagree- The Harmon Curve sounds bad and it's potentially dangerous for good sound

    31.0%
  2. Somewhat Disagree - I like some things about it, but I'd rather not be the target

    43.1%
  3. Somewhat Agree- I like more things about it than not

    19.8%
  4. Strongly Agree - Now, this is how music should be heard!

    0 vote(s)
    0.0%
  5. Fuck me in the ass with the IE 2017 treble boost! (Harmon Curve is for kids with hearing loss)

    6.0%
  1. ChaChaRealSmooth

    ChaChaRealSmooth SBAF Gearmaster

    Staff Member Friend Gearmaster
    Joined:
    Dec 6, 2018
    Likes Received:
    5,931
    Dislikes Received:
    14
    Trophy Points:
    93
    Location:
    Avalon

    Attached Files:

    Last edited by a moderator: Nov 18, 2020 at 8:24 AM
    dncnexus and ultrabike like this.
  2. Lyander

    Lyander Too sensitive for SBAF

    Friend
    Joined:
    Feb 25, 2017
    Likes Received:
    6,989
    Dislikes Received:
    25
    Trophy Points:
    113
    Location:
    Pasig, Philippines
    Self-admitted basshead here so that explains my not hating it entirely— the target voicing hurts a bit from upper mids onwards though. I want to clarify that I still take strong exception to the curve's being referred to as a "neutral" reference regardless of preference because WTF.

    ADDENDUM: in the interests of keeping anyone from calling this thread a hack job can we all acknowledge that the above curves are what measures as coinciding with the Harman Target using the specific setup Sean Olive et al had on hand and shouldn't be taken as universal truth on everybody's EARS/RANDYs/Scrapses/AliExpress GRAS knockoffs regardless of compensation curve magycks invoked, and also that even if it happens to align well with someone's preferences that does not mean it is in any capacity deserving of the "neutral" moniker?
     
    Last edited: Nov 9, 2020
  3. Soups

    Soups Sadomasochistic cat

    Contributor
    Joined:
    Dec 26, 2017
    Likes Received:
    1,015
    Dislikes Received:
    1
    Trophy Points:
    93
    Location:
    Refractory
    Recently acquired the Moondrop Blessing 2, which is supposedly very close to the Harman IE curve. Although it's not how I'd want my speakers to sound, I'm generally pleased with the tonality of the B2's. Tonally, I even prefer it to the FDX1 with any of the barrel options.
     
  4. Vtory

    Vtory Illogical Spock

    Friend MZR
    Joined:
    Sep 12, 2016
    Likes Received:
    6,245
    Dislikes Received:
    39
    Trophy Points:
    113
    Location:
    Greater Philadelphia Area, PA
    As I don't think the graph attached in the OP very informative, I plotted equalized (predicted) frequency responses myself by using REW v5.2-b55, the headphones being fixed (HFM HE6SE).

    For those who are familiar with sbaf-compensation and Minidsp results, the plot below could be easier to imagine their sounding.

    [​IMG]

    Note that the original he6se measurement taken with my pair and specific unit of Minidsp EARS I've been using. SBAF comp(v1)-based equalization was computed with REW. I also used the Harman AE (Around-Ear) 2018 target-based equalization as per the following.

    upload_2020-11-10_3-5-4.png

    Of course, I believe well-seasoned sbafers focus on relative deltas rather than proximity to flat lines. For your reference, after applying the harman eq, the headphones are claimed to have reasonably flat response in Gras and Harman-coordinates. See this doc for more details.

    Here is the breakdown of my overall thoughts by three major frequency regions.
    • 200hz or below : No need for additional introduction. Pick your own poison based on your bass-head-ness.
    • 800 to 5khz : Practically more bothering area than the above (as bass boosting much easier to ignore). I run into a lot of struggles when looking at Harman-compensated results (mostly taken in Gras) these days. At least to me sbaf-target is more in line with my perception of neutrality in upper mids and lower highs.
    • 8k to 10khz : I know there (arguably) exists a fundamental limitation in this region. But as a person who prefer to have a narrow dip at 9khz (in sbaf-coordinates), I might prefer the Harman one here.
     
    Last edited: Nov 10, 2020
  5. spwath

    spwath Collegiate hijinks master

    Friend BWC
    Joined:
    Dec 13, 2015
    Likes Received:
    6,296
    Dislikes Received:
    29
    Trophy Points:
    113
    Location:
    Hartford, CT
    That 64% number doesnt seem to be working out too much among this demographic.
     
  6. Lyander

    Lyander Too sensitive for SBAF

    Friend
    Joined:
    Feb 25, 2017
    Likes Received:
    6,989
    Dislikes Received:
    25
    Trophy Points:
    113
    Location:
    Pasig, Philippines
    I don't think the forum participants (at least the more vocal ones) can rightfully be shoehorned into the "average consumer" demographic.

    Also several people have been publicly critical of the target (cough) so the fact that the votes are publicly visible might inspire some worry of standing out in a bad way.
     
    sphinxvc and nishan99 like this.
  7. lagadu

    lagadu Facebook Friend

    Contributor
    Joined:
    Jun 11, 2019
    Likes Received:
    111
    Dislikes Received:
    0
    Trophy Points:
    33
    Location:
    PT/DK
    I just don't like the upper mids. Bass is generally fine for me but a little less would be OK too.
     
  8. Vtory

    Vtory Illogical Spock

    Friend MZR
    Joined:
    Sep 12, 2016
    Likes Received:
    6,245
    Dislikes Received:
    39
    Trophy Points:
    113
    Location:
    Greater Philadelphia Area, PA
    Regarding the harman target and its underlying studies, there are several common things people (myself once included) often misunderstand. Sean and his team have driven this issue in a very bias-inducing way.
    • Denominator (aka target population, model specification, etc depending on the context) and generalizability matter. The result only generalizable to the population from which the study participants could be proportionally drawn from. Nearly all sort of scientific studies have the same issue, but Sean used to be extremely naive in referring to limitations.
    • ^ means the result AT BEST explains random people drawn from non-hobby space. ONLY in terms of the central tendency. Person-specific predictive models neither derived nor proposed.
    • On top of that, more accurate description of Sean's work I believe is NOT the preferred 'headphone target ' in general BUT the preferred intensity/amplitude of peak/dip (at two fixed frequencies and fixed Q-factors), which added to Harman loudspeaker target (with HATS transformation applied). We are talking about Y in the context of "X is preferred over Y". It makes the study way more doable, but I've never seen Sean clearly pointed this out in any presentation he gave to the general audience.
    • Even within the well-defined boundary of his research scope, the preference experiment always varied across multiple covariates (one key such thing I remember is the base headphones where equalization applied and tested). And he seems to avoid 'fair' comparison (the eq vs non-eq comparison test never performed on tonally-well received headphones).
    • Many technical shortcomings (power, modeling, etc) observed in the study as well if I loosely nitpick as a quantitative researcher. But I'd digress.
     
  9. E_Schaaf

    E_Schaaf MOT: Upscale Audio; recovering HFM addict

    Friend BWC
    Joined:
    Mar 28, 2017
    Likes Received:
    4,756
    Dislikes Received:
    7
    Trophy Points:
    93
    Location:
    Pasadena, CA
    I do like a broad bass boost and a bump or two in the mids. Assuming we've got good dynamic control, I like a flat bass shelf (no localized emphasis) at like +3-6dB <250hz for most music, <400hz for lean tracks, with no more than +1-2db at 1khz. And then tilt the whole thing down a degree or two for the highs. 1khz emphasis only works for me if it's still at or below the level of the upper bass so as not to sound overly forward. Sub-only emphasis in the low end doesn't do shit for most acoustic music. Do 64% of people hate lower mids and upper bass?

    But also how the hell is listening level not accounted for in these surveys? We already know perception of an equal distribution of frequencies shifts quite a bit in the FR extremes as we up and down the volume (Fletcher Munson curves or similar?). The ear's always the most sensitive somewhere in that 1-4khz area which is why a 'BBC dip' can help you crank things up without your ears bleeding. How loud does the average consumer listen?

    I just want to know more about their testing methodology. And I certainly don't think one particular curve should be the 'target' for everything that exists. There must be some degree of feedback going on - headphones that sound a certain way become ubiquitous, so people find themselves listening to music that's flattered by that particular curve, and then recordings get mastered with that curve in mind, and then the curve becomes a 'standard' as opposed to a personal preference. Wonder if this is one of the many factors that plays into the fact that you can tell if a hit track was released in the 60s, 70s, 80s, 90s, 00s, 10s just by the mastering, even if it's a simple arrangement.
     
    Lyander, Jinxy245, Sqveak and 4 others like this.
  10. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    68,025
    Dislikes Received:
    112
    Trophy Points:
    113
    Location:
    Antarctica
    Thanks for doing this! I've added the above to the first post.
     
  11. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    68,025
    Dislikes Received:
    112
    Trophy Points:
    113
    Location:
    Antarctica
    One thing I want to add is that I'm only somewhat opposed to the Harmon target. I get it. When I'm at work and hand people decent headphones to try, I observe that the first they say is "more bass", crank up the bass boost, and maybe later on move up the treble slightly.

    On the poll, I voted strongly disagree.

    The reason for this is the damage that this has caused in terms of misunderstanding what the Harmon target means. For example, I've seen people refer to this curve as a "Harmon-neutral" as if this was a variant of neutral. LOL, it's not! I get why RTings uses the consumer reference curve, but on the other hand, for televisions, RTings goes for studio professional calibration, not Best Buy Vivid mode calibration, which is essentially what the Harmon target is for headphones. The same thing now for ASR, which is getting into the headphones measurements arena. ASR measures speakers against neutrality, with a predictive listening room analysis and reference line for perceptual neutral. However, ASR then uses this consumer preference curve for its headphones.

    When it comes to a target headphone curve, we are better off with something that is closer to perceptual neutral. Not something that is not based on consumer preference, but something that sound engineers in media and entertainment are used to, something that the artists, directors, and producers sign-off in the mastering room or mix stage. Something that is accordance of what we hear in most newer movie theaters. Something that is more similar to what Sonarworks, DIYaudioheaven, and SBAF subscribes.

    At the end of the day, we will probably lose this fight because the Internets is stupid; but at least I know we tried.
     
    Last edited: Nov 10, 2020
    Mont789, 9suns, Kouon and 16 others like this.
  12. Priidik

    Priidik Friend

    Friend BWC
    Joined:
    Sep 27, 2015
    Likes Received:
    1,723
    Dislikes Received:
    3
    Trophy Points:
    93
    Location:
    Estonia
    Industrial city noise 'burns' the sensitivity of the ear and Harman curve attempts to compensate.
     
    Lyander likes this.
  13. nishan99

    nishan99 Almost "Made"

    Joined:
    Dec 1, 2019
    Likes Received:
    285
    Dislikes Received:
    6
    Trophy Points:
    63
    Location:
    Layla
    The Harman curve bass boost is to totally a preference thing. Neutral bass is flat measuring bass to my ears. If the headphones/speakers have a wimpy bass then an elevation there can be more satisfactory and even neutral.

    The Harman curve upper mids and upper treble is my second criticism, the upper mids are a bit forward and the upper treble is a bit subdued.

    I am no scientist, just reporting what I am hearing :)
     
  14. Biodegraded

    Biodegraded Friend

    Friend
    Joined:
    May 28, 2017
    Likes Received:
    4,235
    Dislikes Received:
    1
    Trophy Points:
    113
    Location:
    Vancouver BC
    To @Lyander 's point about the specific setup, I think it's worth distinguishing between the headphone and IEM targets. I'm less used to listening to Harman-targeted headphones than IEMs, but looking at @Vtory 's graph I think I'd have fewer issues with the c. 2 dB elevation of the headphone Harman vs SBAF upper mids than I know I have with that area in the IEM curve (which is often measured on cheap IEC711 knockoff couplers that differ quite a bit from the (both fake and real) GRAS units - yet another source of variance).

    And more bullets for Vtory's list: the listening sessions at different FR settings in the Harman tests are by necessity short, so any fatigue factor is reduced; and because they're comparative (the 'X preferred over Y' thing), the psychoacoustically prominent areas of bass and upper mids would tend to be uppermost in listener consciousness.
     
  15. perogie

    perogie Facebook Friend

    Joined:
    Aug 16, 2020
    Likes Received:
    102
    Dislikes Received:
    0
    Trophy Points:
    33
    Location:
    Ontario
    One of my biggest issues with the curve is it’s presentation to the general public. Rarely, if ever, do you see any context provided for its use or why it is even used. It can be very misleading to reference something as a baseline representation of “correct” when it clearly isn’t. Last time I looked into it only 2/3 of whatever random people preferred the curve, that’s far from the universal backing required to be considered a reference in my opinion.
     
    Josh83, memoryerror, dncnexus and 4 others like this.
  16. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    68,025
    Dislikes Received:
    112
    Trophy Points:
    113
    Location:
    Antarctica
    ^ Exactly. It's how the target is being represented and also the lack of clarification from Dr. Olive in regards to popular interpretation and understanding of the target. I would surmise that it's not in Olive's best interest to clarify. As an academic who write papers for AES, it's not in his best interest. The more attention he can get, the more widespread the Harmon target, the more he is able to hold tenure at Harmon.
     
    Cryptowolf, nishan99 and perogie like this.
  17. Vtory

    Vtory Illogical Spock

    Friend MZR
    Joined:
    Sep 12, 2016
    Likes Received:
    6,245
    Dislikes Received:
    39
    Trophy Points:
    113
    Location:
    Greater Philadelphia Area, PA
    I do think it rather a better case. Harman team consistently uses the word "accurate" (as defined by the proximity to their target) in their manuscripts and presentations. For example, one title was "Some new evidence that teenagers and college students may prefer accurate sound reproduction".. I am not joking, it was an aes paper... He is a big ego in this field, and more influential than he deserves. At worst it may facilitate lots of misleading decisions.

    And pretty unfortunately, it is not a well-funded research area where active cross-criticism/validations occur academically. State/federal governments may have zero interest to offset the induced bias.
     
  18. purr1n

    purr1n Finding his inner redneck

    Staff Member Friend BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    68,025
    Dislikes Received:
    112
    Trophy Points:
    113
    Location:
    Antarctica
    That could have been a different paper in respect to The SR009? EQ'd to certain profiles (not necessarily Harmon back then) that fit certain headphones in existence. Keep in mind that in 2010-2012, there were a lot of crappy headphones with huge treble peaks.

    To his credit, Dr. Olive did do a lot of work, and IME with most regular consumers (that is outside of audiophiles, pros or even amateurs producing content), the curve does fit their preferences.

    My opinion that that the last piece in the playback system should be neutral sounding, and that preference curves should be managed by playback software or EQ.
     
    Last edited: Nov 10, 2020
  19. Biodegraded

    Biodegraded Friend

    Friend
    Joined:
    May 28, 2017
    Likes Received:
    4,235
    Dislikes Received:
    1
    Trophy Points:
    113
    Location:
    Vancouver BC
    Context. That paper refers to loudspeakers and to lossy vs lossless files, neither of which are being discussed here, and which I suspect there'd be less objection to here.
     
  20. spwath

    spwath Collegiate hijinks master

    Friend BWC
    Joined:
    Dec 13, 2015
    Likes Received:
    6,296
    Dislikes Received:
    29
    Trophy Points:
    113
    Location:
    Hartford, CT
    The meeting I was in with Mr. Olive that sparked this discussion, he seemed pretty honest about what the curve was. He said it was designed to please 64% of the population, never used the word neutral or anything.
     
    9suns, Kouon, Josh83 and 8 others like this.

Share This Page