Headphones and the Harman target curve

[Phos]

The problem I've run into with regards to evaluating the HTR is figuring out how close my headphones are to start with, which I'm kind of on my own with because my headphones are HiFiMan HE-500's with Lawton Audio pads. My attempt to just put the HTR curve into my parametric EQ has me involuntarily trying to unseal my Eustachian tubes, and even reducing the suck out it still sounded odd, so something isn't as it seems. Just compensating for it relative to the old innerfidelity raw measurements has the effect of also smoothing out a couple of notable dips in the FR so it isn't the only significant change being made.

Somewhat later edit: Seems my take years ago, on my ears at least, that the Lawton Audio pads sounded like the old fashioned velour pads was correct, using a sine generator program on my computer I was able find the same dips and peaks.

 

[Theriverlethe]

Read the anechoic FR graph - total variation from flat is approx. 8 dB. That's a decent bit - not exactly a benchmark neutral system when digital room correction exists.

I also am not the person you responded to originally, so I never even said that.

Did you bother to read this paragraph?

"How these peaks are perceived will also depend on the speaker's dispersion. Fig.5 shows the lateral dispersion, normalized to the response of the tweeter axis. Although B&W claims that its large-diameter midrange unit has wide dispersion, you can see from this graph that the midrange driver becomes quite directional above 1kHz. In all but small rooms, this will work against the audibility of the small presence-region peak in the on-axis response, meaning that the 800 Diamond will sound more neutral in this region than is suggested by the on-axis measurement. Similarly, what appears to be a narrow off-axis flare at 5.3kHz is actually a small on-axis suckout filling in to the speaker's sides. Finally, this graph shows that the tweeter is a little more directional in the top audio octave than is typical for a 1" diaphragm; again, this will ameliorate the effect of the on-axis boost in the same region in all but small rooms."

 

[Robbo99999]

I thought I'd type my quick findings on how a person might change the overall tone of the Harman Curve to their liking - a "Tilt Control". I've worked out a filter that can be put into EqualiserAPO (for instance) on top a Harman Curve EQ to then tweak the overall tone in a very broad manner. Here it is in action with the pic just showing the effect of the filter on a frequency response of a headphone (happens to be HE4XX):

You can see it tilts the frequency response in an almost straight line from around 100Hz to 20,000Hz, which you can see from the red line I drew on top of the shaded turquoise area above (the shaded area is showing the how the filter effect changes as you go up the frequency range - essentially a straight line, which is what you'd want from a tone control). The filter to use is High Shelf Filter at 982Hz Q0.2, and this is essentially a tilt filter, tone control. I have to admit I've not tried using this on top of the Harman Curve yet, but it could be useful in tuning an existing Harman Curve EQ to the different tonalities that some albums might be recorded at.....some albums are too bright, some too dull, you could use this filter as a broad tone control. I've yet to try it, but thought I'd post it here.

EDIT: I've experimented with my tone control, and I find it very effective. For instance, Red Hot Chili Pepper songs I find are generally recorded too bright, so my tone control filter fixes that with just -1 or -2dB on the Gain for that filter, you could probably even try it with just 0.5dB increments. You can easily notice 1dB changes in that filter because it tilts the whole frequency response from 100Hz upwards. High Shelf at 982Hz, Q 0.2 -> that's the filter I urge you to give it a try as a tone control. (that's naught point two for the Q filter, it's a very low Q filter!).

 

[preload]

I think we can put to bed the ‘science’ bit.

Preference isn’t science.

Measurements are science.

You can measure preference scientifically.

That doesn’t make the preferences scientific.

Wow can’t believe I missed this. It makes absolutely no sense.
The measurement of listener preferences as part of an experiment can absolutely be considered “science.”

 

[JJB70]

If people like the Harman curve then like it. If people don't like it then don't like it. It's a preference curve, that it may represent a response most like is irrelevant if you don't like it. Preferences for FR curves are one of those things where people with one are no more right or wrong than those with a preference for something else.
There seems to be two increasingly prevalent assumptions in the world of headphones:

-The Harman target is THE curve rather than a curve, with alternatives being 'wrong'; and
-Liking a particular response means people don't like others

I have nothing against the Harman target and quite like it for pop music but I also like other options and for classical music find the bass is overdone on Harman tuning.

 

[Sean Olive]

Wow can’t believe I missed this. It makes absolutely no sense.
The measurement of listener preferences as part of an experiment can absolutely be considered “science.”

Yes saying measuring listener preferences is not scientific makes no sense, and no arguments are made to support it.

If you can make a hypothesis (e.g. most people prefer flat loudspeakers), test it in a controlled way that produces reliable/valid results and draw conclusions from the evidence then its a science. The fact that humans are the measurement devices make it no less a science than the science of measuring humans' preferences for food and wine (for which many academic books, universities and corporate labs are dedicated)


The fact that you can model and predict those preferences based on acoustic measurements of the loudspeakers/headphones indicates that there is a strong relationship between the physical properties of the devices, the sounds they produce, and how humans perceive them. This is something I'm not sure can be done as reliably with food and wine.

 

[Sombreuil]

Wow can’t believe I missed this. It makes absolutely no sense.
The measurement of listener preferences as part of an experiment can absolutely be considered “science.”

Science ≠ correct/true.
Science = trying to push knowledge.

Your definition of "science" is wrong in the first place.

 

[Feelas]

Science ≠ correct/true.
Science = trying to push knowledge.

Your definition of "science" is wrong in the first place.

Science = using the scientific methods and pretty much that's it. How can a study using purely scientific, statistical methods be called unscientific is beyond me.

If people like the Harman curve then like it. If people don't like it then don't like it. It's a preference curve, that it may represent a response most like is irrelevant if you don't like it. Preferences for FR curves are one of those things where people with one are no more right or wrong than those with a preference for something else.
There seems to be two increasingly prevalent assumptions in the world of headphones:

-The Harman target is THE curve rather than a curve, with alternatives being 'wrong'; and
-Liking a particular response means people don't like others

I have nothing against the Harman target and quite like it for pop music but I also like other options and for classical music find the bass is overdone on Harman tuning.

Yet if some curve is somehow closer to natural sound perception, then it is not merely a preferential curve and people should get that into the system. One reproduction curve can be closer to real hearing that other ones and it's futile to say otherwise.

 

[Theriverlethe]

Science ≠ correct/true.
Science = trying to push knowledge.

Your definition of "science" is wrong in the first place.

I'd say it's closer to "trying to attain knowledge."

 

[Yorkshire Mouth]

Yes saying measuring listener preferences is not scientific makes no sense, and no arguments are made to support it.

If you can make a hypothesis (e.g. most people prefer flat loudspeakers), test it in a controlled way that produces reliable/valid results and draw conclusions from the evidence then its a science. The fact that humans are the measurement devices make it no less a science than the science of measuring humans' preferences for food and wine (for which many academic books, universities and corporate labs are dedicated)


The fact that you can model and predict those preferences based on acoustic measurements of the loudspeakers/headphones indicates that there is a strong relationship between the physical properties of the devices, the sounds they produce, and how humans perceive them. This is something I'm not sure can be done as reliably with food and wine.

With all due deference to your clear superiority in the field of audio, I will reply shortly as to why LOGICALLY and SCIENTIFICALLY this is an inappropriate term.

In the meantime, and to underline the point, I am still extremely appreciative of the excellent work you’ve done.

For me, the one thing has nothing to do with the other.

 

[DeLub]

One reproduction curve can be closer to real hearing that other ones and it's futile to say otherwise.

Due to the “circle of confusion” there is no single one FR that comes closer to real hearing in all circumstances. It may only be possible to say that a specific FR is the best, or the most natural sounding, for a specific recording situation.

 

[Chromatischism]

Science ≠ correct/true.
Science = trying to push knowledge.

Your definition of "science" is wrong in the first place.

Did you stumble onto the wrong forum?

 

[JJB70]

Yet if some curve is somehow closer to natural sound perception, then it is not merely a preferential curve and people should get that into the system. One reproduction curve can be closer to real hearing that other ones and it's futile to say otherwise.

Perception of others and hearing of others is not necessarily my perception or hearing. I would say it is futile to imagine any statistical analysis of preference is proof either that a curve is closer to real hearing (what is real hearing?) or closer to natural sound perception. I might agree it sounds quite natural for pop music, I don't think it is natural for classical music as the boosted bass sounds unnatural to me.

 

[Feelas]

Due to the “circle of confusion” there is no single one FR that comes closer to real hearing in all circumstances. It may only be possible to say that a specific FR is the best, or the most natural sounding, for a specific recording situation.

Not quite, since you can trace a curve w/ minimal difference over whole FR for thousands of recordings and use that as a reference point. One could try proving that discrepancy due to (overused here) "circle of confusion" is so big that tracing one curve is impossible, but yet nobody dared to, it seems and the discussion is done on what-if manner, instead of actual evidence. I'd be very reluctant to call Harman further from truth than most headphones which have overall a dramatic bass shortage below 70Hz.

Perception of others and hearing of others is not necessarily my perception or hearing. I would say it is futile to imagine any statistical analysis of preference is proof either that a curve is closer to real hearing (what is real hearing?) or closer to natural sound perception. I might agree it sounds quite natural for pop music, I don't think it is natural for classical music as the boosted bass sounds unnatural to me.

It is quite futile to imagine one universal curve (especially since we've seen posts detailing research regarding actual ear canal resonance, so we can interpolate that ear shape might also have variation), yet as mentioned above: one can trace a line being the closest possible to thousands or hundreds of thousands of records. And perhaps it's not very sincere to assume that one is a very special person. Considering that one has pretty average anatomical featues would probably work well to remind oneself that perhaps the averaged rating might happen to be what one will like. But hey, how would you sell worse stuff if people didn't assume that they're special?

FYI classical music is never bass short, especially if you're in the auditorium a few meters from the orchestra. B&K research showed that treble roll-off is perceived as natural and neutral. I cannot see why people keep rationalizing that bright sound does justice to music. Bass range is incredibly important!

As to what is mentioned regarding statistics and yet nobody pursues to argue with that: Floyd Toole has shown that in blind testing people don't stray from ranking objectively better speakers with higher ratings, and the effect holds whenever you trim the ranges (you remove the best and worst speakers) as the relative preference is kept. One can thus extrapolate, that if for speakers a statistically important group chooses neutral and accurate as better, that for headphones the same will hold. What holds in blind testing is more important, since with that one can push the technological envelope instead of merely selling something different, albeit never better.

 

[Sombreuil]

Did you stumble onto the wrong forum?

No but I did answer to the wrong person though.

 

[JJB70]

Not quite, since you can trace a curve w/ minimal difference over whole FR for thousands of recordings and use that as a reference point. One could try proving that discrepancy due to (overused here) "circle of confusion" is so big that tracing one curve is impossible, but yet nobody dared to, it seems and the discussion is done on what-if manner, instead of actual evidence. I'd be very reluctant to call Harman further from truth than most headphones which have overall a dramatic bass shortage below 70Hz.


It is quite futile to imagine one universal curve (especially since we've seen posts detailing research regarding actual ear canal resonance, so we can interpolate that ear shape might also have variation), yet as mentioned above: one can trace a line being the closest possible to thousands or hundreds of thousands of records.

FYI classical music is never bass short, especially if you're in the auditorium a few meters from the orchestra. B&K research showed that treble roll-off is perceived as natural and neutral. I cannot see why people keep rationalizing that bright sound does justice to music. Bass range is incredibly important!

I agree bass is important and am familiar with live performance of classical music. However the prevalence of boosted bass means many now interpret an absence of boosted bass as being bass light. I disagree, and find the tuning of my Etymotic ER4SR much closer to a natural sound of classical music.

 

[Feelas]

I agree bass is important and am familiar with live performance of classical music. However the prevalence of boosted bass means many now interpret an absence of boosted bass as being bass light. I disagree, and find the tuning of my Etymotic ER4SR much closer to a natural sound of classical music.

Perhaps the popular preference might be skewed towards bass-heavy, albeit if we skip the argument from popularity then that's pretty much an assumption not really reinforced by any consistent research, I think.

Touching on classical music, I have to ask: which exact concert hall would you consider to be the "real" one? Spending merely a minute on googling shows a bunch of papers which highlight that there's not even a consensus on how a concert hall sounds, much less what constitutes a real performance. And as I mentioned: that aspect can be addressed by getting a FR which is statistically the closest to any possible variation on real cases...

 

[JJB70]

Perhaps the popular preference might be skewed towards bass-heavy, albeit if we skip the argument from popularity then that's pretty much an assumption not really reinforced by any consistent research, I think.

Touching on classical music, I have to ask: which exact concert hall would you consider to be the "real" one? Spending merely a minute on googling shows a bunch of papers which highlight that there's not even a consensus on how a concert hall sounds, much less what constitutes a real performance. And as I mentioned: that aspect can be addressed by getting a FR which is statistically the closest to any possible variation on real cases...

I quite agree that concert venues differ greatly, performances also differ. Which of course is part of the joy of live music. But different people have different perceptions and it is an inherently subjective and personal thing.

 

[Feelas]

But different people have different perceptions and it is an inherently subjective and personal thing.

Of course I take that into account. The range of discrepancy itself can be measured and the subjective & personal perception will probably not behave chaotically outside of some specific range. After all there are not many people on the planet (I think?) who have half a meter ears.

 

[Theriverlethe]

Of course I take that into account. The range of discrepancy itself can be measured and the subjective & personal perception will probably not behave chaotically outside of some specific range. After all there are not many people on the planet (I think?) who have half a meter ears.

People seem to forget that the Harman research actually specifies a range. "Expert listeners" prefer less bass and the uninitiated prefer more, so the Harman target is a middle ground that makes no one happy.