Not sure I like Harman curve

[MayaTlab]

I have not. I have ordered K371s so I can compare.

Good idea to try other headphones tuned to the target. The thing is, there's always going to be a bit of leeway between :
- your own preferences (statistically speaking not anything goes however, most people tend to prefer headphones trending towards reasonably shaped targets),
- sample variation, which can either be severe or not a problem at all depending on the model,
- how specific models "couple" with your head.
The latter two mean that it isn't straightforward to be quite certain that what you're hearing really is what the Harman target is meant to sound like. The K371 among other foibles is quite susceptible to leakage and may not deliver, once on your head, the bass response you'll see on an ear simulator.

 

[Cars-N-Cans]

See the link in post 12 and the paper of Sean Olive linked therein. At least read the conclusion. It’s easily written.

Not the targeted recipient but thanks for the link.

As Amir mentions the curve was referenced to preferences for loudspeakers, which at first caught me as a bit of a circular argument when I first heard it mentioned. A bit like Harman saying if you like our speakers, you will love our headphones! But with respect to a nearfield system I referenced in my past post, if I look past the natural differences in imaging, the perceived tonality is identical for all intents and purposes between the speakers and headphones in A/B comparisons. Any differences I do note, correlate very closely with remaining deviations from the Harman curve despite both systems being calibrated completely differently (nearfield speakers by me, headphones via mfr. and available measurements). But I guess if there was a difference then we would be in trouble as we would still not know what constitutes tonally neutral (or more likely I would just be wrong in my measurement methods ). But if I like the result then I guess I cannot complain. Well actually I suppose I could since I want companies like GRAS to make their fixtures cheaper so I can measure my headphones like the pros do. Oh well...

 

[Jimbob54]

BTW, If Harman is the way of the future why do we get so upset about DAC filters that cut off at 19k instead of 20k. There is nothing to hear anyway, equalising to a Harmon curve starts to kill everything above 15k!

People still have full range speakers

 

[nyxnyxnyx]

I bought akg371 based on hyper-positive experiences from what I can read on the internet. Heard it for a few weeks and I found it kinda lame and especially lacking slam and impact. Tried EQ to multiple profiles/targets but still do not like it that much.
But I don't blame the target, I think it's just not for me, and after all, we all are responsible to find something we truly like instead of sticking to something we don't due to some sort of validations.

 

[HarmonicTHD]

Not the targeted recipient but thanks for the link.

As Amir mentions the curve was referenced to preferences for loudspeakers, which at first caught me as a bit of a circular argument when I first heard it mentioned. A bit like Harman saying if you like our speakers, you will love our headphones! But with respect to a nearfield system I referenced in my past post, if I look past the natural differences in imaging, the perceived tonality is identical for all intents and purposes between the speakers and headphones in A/B comparisons. Any differences I do note, correlate very closely with remaining deviations from the Harman curve despite both systems being calibrated completely differently (nearfield speakers by me, headphones via mfr. and available measurements). But I guess if there was a difference then we would be in trouble as we would still not know what constitutes tonally neutral (or more likely I would just be wrong in my measurement methods ). But if I like the result then I guess I cannot complain. Well actually I suppose I could since I want companies like GRAS to make their fixtures cheaper so I can measure my headphones like the pros do. Oh well...

You misunderstood. Read the paper. It is not about Harman speakers. It is about that Toole based on many blind listening studies (ABX) found that on average people prefer speakers which measure anachoically flat on axis and have smooth directivity. Now based on that Olive took this insight one step further and applied it to headphones. As headphones sit close or even in ear the frequency response is not flat anymore (as with the speakers) because the influence of your outer ear must be compensated for. Olive again verified it with ABX studies. Details see the linked paper.

In summary as Amir also stated - the “wavy” headphone curve is the equivalent to the flat speaker curve.

 

[Cars-N-Cans]

You misunderstood. Read the paper. It is not about Harman speakers. It is about that Toole based on many blind listening studies (ABX) found that on average people prefer speakers which measure anachoically flat on axis and have smooth directivity. Now based on that Olive took this insight one step further and applied it to headphones. As headphones sit close or even in ear the frequency response is not flat anymore (as with the speakers) because the influence of your outer ear must be compensated for. Olive again verified it with ABX studies. Details see the linked paper.

In summary as Amir also stated - the “wavy” headphone curve is the equivalent to the flat speaker curve.

My apologies as I was being a bit tongue-in-cheek with that comment, so please don't take it too literally. My comment was in regards to the fact that Floyd Toole's research shows that we like speakers that are flat on-axis and have good directivity, and extrapolating this to headphones struck me at the time as a bit of a tautological argument philosophically. Technically I have no objections, and the fact that there is excellent correlation between my preferences for speakers and headphones shows that its completely valid, which put my philosophical "concerns" at ease, at least for me. Even before reading the paper I understood well why the curve has the shape it does, namely the characteristics of our ears with respect to anatomy and acoustics. Perhaps maybe my viewpoint is from working too many years at a research facility with highly regarded physicists who always found interesting ways to poke holes and deflate one's argument in discussions.

One could still argue that how do we know that what we consider "good" is really good? Speakers are tonally neutral, but with respect to imaging they are not the same as what one would hear in person. How does that impact our perception of them? If we had an alternative that could be completely accurate and reproduce the acoustic sound field verbatim, how would we then view regular loudspeakers and headphones (that lack HRTFs despite having IIDs and ITDs) once we have that knowledge? I'm sure studies have been done, but still its a question I think about.

At any rate, I don't want to get too far off topic, but getting back to the Harman curve and the topic of this thread, I wonder how many people it actually is subjectively acceptable for? Even though I'm sure their data would provide some good metrics, quantifying things that are subjective is difficult to say the least. Nobody says you have to use it, but there is also the issue that one doesn't want to acquire a taste for something improper merely because its what's viewed as orthodox. Not really stating these as absolutes or that the target is wrong, more along the lines of open questions for pondering. (Edit: Correcting bad spelling.)

 

[MayaTlab]

I bought akg371 based on hyper-positive experiences from what I can read on the internet. Heard it for a few weeks and I found it kinda lame and especially lacking slam and impact. Tried EQ to multiple profiles/targets but still do not like it that much.
But I don't blame the target, I think it's just not for me, and after all, we all are responsible to find something we truly like instead of sticking to something we don't due to some sort of validations.

I'll sound like a broken record, but the K371 is not appropriate to know what the Harman target sounds like. Given how it varies at lower frequencies across human subjects (https://www.rtings.com/headphones/1-5/graph#1671/7913), it's already partly immaterial how it measures on an ear simulator (without even taking into account sample variation and non-leakage related coupling issues).

I actually don't think that a singular pair of headphones EQed to the target (or targeting it in the first place) is enough to judge whether one likes it or not. I believe that averaging how several headphones, EQed to the target (even better if it's according to measurements of your own samples on a compatible industry standard), measure in situ, and then EQing an individual pair of headphones to these averages is a better approach, but even then it isn't straightforward : https://audiosciencereview.com/foru...ctivist-bang-for-your-buck.34596/post-1207611

 

[MayaTlab]

At any rate, I don't want to get too far off topic, but getting back to the Harman curve and the topic of this thread, I wonder how many people it actually is subjectively acceptable for?

It's quite unfortunate that so much of the actual content is behind a paywall as most of your questions are, to a certain degree, provided some form of answer in them .

For the over-ears target, within a certain degree of uncertainty, that unfortunately likely is above minimum threshold of audible differences (an illustration of that here : https://audiosciencereview.com/foru...ew-rode-nth-100-headphones.32296/post-1169984), quite a few elements suggest that it's a pretty solid guideline, such as :
- it was mostly tested against alternative curves on large, open dynamic headphones, which have been demonstrated to be quite leakage tolerant, fairly consistent across individuals, and with a decent match between individuals and ear simulator up to a few kHz (it's for this type of headphones that we have the most significant amount of on-head behaviour research I believe)
- it was tested on several such headphones, alleviating issues related to different physical formats coupling differently with different users
- it was tested "blind" against alternative curves (the listeners didn't know they were listening to the target vs an alternative)
- some of the alternative curves it was tested against were quite close to it, which could have indicated possible preferential trends around the target to slightly nudge it one way or another as it was developed (unlike for the in-ears where other than one all other alternative curves were far off the HT)
- for some of the studies the listeners panel was quite large
- care was given to assess differences between them such as age
The list could probably go on.

 

[kchap]

People still have full range speakers

True. Given I now know the Harman curve for headphones is supposed to the equivalent of speakers in a Semi-Reflective Field, does that mean the speakers also roll off? Maybe the approach taken by @_thelaughingman to HP EQ should be applied to speakers.

 

[kchap]

See the link in post 12 and the paper of Sean Olive linked therein. At least read the conclusion. It’s easily written.

Thanks.

 

[Jimbob54]

True. Given I now know the Harman curve for headphones is supposed to the equivalent of speakers in a Semi-Reflective Field, does that mean the speakers also roll off? Maybe the approach taken by @_thelaughingman to HP EQ should be applied to speakers.

Harman curve for loudspeakers

I’ve been pondering the harman curve this morning, or more specifically why the convention is to measure headphones with references to the harman curve, but to measure loudspeakers with reference to an ideal flat response. Perhaps for professional monitoring a flat loudspeaker response is ideal...

www.audiosciencereview.com

I would have thought you want a DAC that has flat FR through the listening frequencies then get the speaker response to what is required either with digital DSP before the DAC (non destructive) and/ or through positioning and room treatment. I think people get more upset with DACs with slow filters that start to drop somewhere below 19k and (more importantly) dont drop like a stone after 22. Dont think anyone gets too upset about a steep filter that drops hard at 19k

 

[kchap]

Harman curve for loudspeakers

I’ve been pondering the harman curve this morning, or more specifically why the convention is to measure headphones with references to the harman curve, but to measure loudspeakers with reference to an ideal flat response. Perhaps for professional monitoring a flat loudspeaker response is ideal...

www.audiosciencereview.com

I would have thought you want a DAC that has flat FR through the listening frequencies then get the speaker response to what is required either with digital DSP before the DAC (non destructive) and/ or through positioning and room treatment. I think people get more upset with DACs with slow filters that start to drop somewhere below 19k and (more importantly) dont drop like a stone after 22. Dont think anyone gets too upset about a steep filter that drops hard at 19k

Yes indeed. I was being a little light hearted when I wrote that comment. Then you pointed me to the post by @andreasmaaan.
Many a true word is spoken in jest ...

 

[nyxnyxnyx]

I'll sound like a broken record, but the K371 is not appropriate to know what the Harman target sounds like. Given how it varies at lower frequencies across human subjects (https://www.rtings.com/headphones/1-5/graph#1671/7913), it's already partly immaterial how it measures on an ear simulator (without even taking into account sample variation and non-leakage related coupling issues).

I actually don't think that a singular pair of headphones EQed to the target (or targeting it in the first place) is enough to judge whether one likes it or not. I believe that averaging how several headphones, EQed to the target (even better if it's according to measurements of your own samples on a compatible industry standard), measure in situ, and then EQing an individual pair of headphones to these averages is a better approach, but even then it isn't straightforward : https://audiosciencereview.com/foru...ctivist-bang-for-your-buck.34596/post-1207611

Fair point, but for what it's worth I have multiple pairs of headphones and I all tried to EQ them to the provided target(s). Some of them I like better than the rest, I am not sure why but I can't prove it anyway because there's no way I can do a proper blind test with the headphones I am so familiar with.

Like some other folks commented in this thread, nowadays I will try to EQ it to the target, then if I don't like it I will try to make adjustments. Maybe if there's a chance for me to experience the complete, purest "harman experience" I will consider it, but my experience so far is I like it just not too much.

 

[spartaman64]

i think its more a problem with the EQ. if you are using one of those autoeq profiles i find they almost always kill technical performance and make it sound thin to me. I think headphones aren't meant to have like 20 different filters to try and change their sound signature.

 

[Keith_W]

I am not a fan of the Harman curve either. For me it is a starting point for EQ. For me personally it has too much bass and too much treble, so I cut down on both. Others have mentioned a few reasons, as I see it:

1. The Harman curve is an average. No harm in falling outside the average. Your preferences are different. No harm with that!
2. The frequency response of the headphones on a standardized test rig may be different to the actual frequency response on your head.
3. Your hearing may be non-linear.
4. You may have a collection of recordings which are recorded in a certain way, which causes you to favour certain types of sound.
5. Years of listening to a certain type of sound has caused you to preference that type of sound as "normal".

 

[Cars-N-Cans]

I am not a fan of the Harman curve either. For me it is a starting point for EQ. For me personally it has too much bass and too much treble, so I cut down on both. Others have mentioned a few reasons, as I see it:

I think one reason I do like the Harman curve is I have rather petite ears despite being over 6’ tall. One discovery I made when going to the objectivist side of audio is that my hearing skews quite a lot towards the high frequency spectrum, with my peak sensitivity being above 4 kHz, and it doesn’t start to roll off until I’m above about 13-14 kHz. I’m almost completely insensitive to anything below 30 Hz. The treble region is definitely the area I notice as being lacking the easiest. Bass is more variable. Those would potentially predispose me to likening it and finding it to be very neutral.

Personal preferences will come in as well, obviously, but I would definitely think there will be people who might not like it If their hearing is a certain way. From looking at their paper, the bass seems to be the most variable part, and years of being a bass head would probably put me more towards the upper part of the spectrum, but of note is the Harman curve is right at the limit of what I like bass-wise. More than that is too much.

 

[threni]

i think its more a problem with the EQ. if you are using one of those autoeq profiles i find they almost always kill technical performance and make it sound thin to me. I think headphones aren't meant to have like 20 different filters to try and change their sound signature.

The headphones don't have any filters - that's upstream in the chain, usually at source. It's no different to applying the same filters directly to the flac file. It's only going to sound thin if you cut frequencies which are responsible for...uh...thickness? If you don't like that, don't do it!

 

[spartaman64]

The headphones don't have any filters - that's upstream in the chain, usually at source. It's no different to applying the same filters directly to the flac file. It's only going to sound thin if you cut frequencies which are responsible for...uh...thickness? If you don't like that, don't do it!

yep which is why i dont use autoeq and it might be what the OP is experiencing if they do

 

[Soandso]

I have reservations about how people may apply EQ to speakers/headphones. My impression is the accepted initial step is to microphone measure the different transduced frequencies and then, based on that data pattern, make EQ adjustments (such as those based on the Harman curve, or guidelines for certain kinds of music).

What my left and right aged ears' testing revealed is there are differences how loud (dB) specific frequencies are being heard, but also differences in the pattern of loudness being heard at said frequencies. In other words: in frequencies that are heard at worse than the clinical norm (25dB) the extra dB needed to hear that specific frequency is not always the same for left and right ears.

Then too, among the frequencies where hearing is less than normal the dB required to hear those frequencies does not follow the same pattern in both left and right ears. In one ear the trend of relative decline of frequency hearing does not occur in one ear at 1 kHz, yet for the same 1 kHz frequency in the other ear there is hearing decline.

And furthermore, even though the general trend of worse hearing is occurring in both ears at 2kHz in one ear at subsequent higher frequencies my impaired hearing is still better than it is at 2kHz in the other ear; and hearing falls from there in a pattern of decline at yet higher frequencies. To further demonstrate variability: in one ear there is a different subsequent higher frequency of those beyond 2kHz where I hear better than what that ear's 2kHz pattern of impairment is.

EQ adjustments based on microphone readings with the end goal of giving one a desired sound spectrum seems to be relevant for those with perfect hearing. We can EQ from those readings and yet not be on target if one ear has limited hearing of several specific frequencies at different base-line dB, nor on target for the other ear if it, in turn, has limited hearing of a different set of specific frequencies at different base-line dB.

Hearing loss is evidently different than the Fletcher-Munson "equal loudness" curve. Although clinical evaluation of hearing loss is usually only done from 25Hz to 8kHz the isolation of frequencies tested rules out any artifacts of masking. If you are 65 years of age or older treat the balance dial as your friend and consider whether EQ-ing to personal taste by ear may be preferable to EQ-ing from measurements.

 

[Cars-N-Cans]

I have reservations about how people may apply EQ to speakers/headphones. My impression is the accepted initial step is to microphone measure the different transduced frequencies and then, based on that data pattern, make EQ adjustments (such as those based on the Harman curve, or guidelines for certain kinds of music).

What my left and right aged ears' testing revealed is there are differences how loud (dB) specific frequencies are being heard, but also differences in the pattern of loudness being heard at said frequencies. In other words: in frequencies that are heard at worse than the clinical norm (25dB) the extra dB needed to hear that specific frequency is not always the same for left and right ears.

There is also a differential preference between the ears even with people with normal hearing. If I recall, the right ear tends to be biased towards the treble.

But one problem with headphones is they are not as easy to measure. You can obviously "roll your own", but then it has to be calibrated to be accurate. With speakers its easier since there is less to worry about, albeit there are still nuances there that can trap the unwary. But at least there are COTS microphones and interfaces available for making the measurement that are not too expensive. I have ARTA and REW for speaker measurements, but with headphones I'm pretty much stuck with whatever is posted along with seat of the pants adjustments.