Why does headphones measurement graphs always gets squiggly-er in the treble?

[MrBlitzpunk]

I'm sorry if this is not the right place to ask this question as I'm quite new to this forum.
But i wanted to delve a little bit deeper in the more scientific/physical side of the audio world.

As the title said, why does headphones measurements always gets janky in the upper treble? I also noticed that IEM measurements is a lot smoother in the treble region. Is this a quirk specific to the transducers/build of headphones or is it the limitation of the measurements rig itself?

A second, somewhat unrelated question; less about audio and more about how soundwave works in general. So how does sound wave interacts with space? I noticed that when i play my speakers inside a closed room, i can still hear most of the upper frequency from outside of the room but the bass is significantly reduced. Meanwhile when there's a concert somewhere near my area, i can mostly hear the bass from far away, but not the treble. What is the physical explanation of this?

 

[MrBlitzpunk]

Actually, a 3rd question; i heard that headphones target curves has a upward slope above 1k, to make it sound 'natural' since in real life human ears naturally boost frequencies in 1k-5k range. But if this were the case, shouldn't 'neutral' target curve be inverted and have downward slope above 1k to be truly 'flat'?

 

[HarmonicTHD]

I'm sorry if this is not the right place to ask this question as I'm quite new to this forum.
But i wanted to delve a little bit deeper in the more scientific/physical side of the audio world.

As the title said, why does headphones measurements always gets janky in the upper treble? I also noticed that IEM measurements is a lot smoother in the treble region. Is this a quirk specific to the transducers/build of headphones or is it the limitation of the measurements rig itself?

A second, somewhat unrelated question; less about audio and more about how soundwave works in general. So how does sound wave interacts with space? I noticed that when i play my speakers inside a closed room, i can still hear most of the upper frequency from outside of the room but the bass is significantly reduced. Meanwhile when there's a concert somewhere near my area, i can mostly hear the bass from far away, but not the treble. What is the physical explanation of this?

Simply speaking. The headphone target curves are derived by Olive, S (former Harman researcher) from the (Harman) speaker preference curves (anechoically flat) less the „filtering“ the outer ear provides. You can google S Olive and get tons of more details than I am able to provide here.

 

[solderdude]

why does headphones measurements always gets janky in the upper treble? I also noticed that IEM measurements is a lot smoother in the treble region. Is this a quirk specific to the transducers/build of headphones or is it the limitation of the measurements rig itself

Its the pinna and earcanal that make things more squigly above 6kHz. The idea is that when one emulates 'average' pinnae and ear canals it is more representative to what we hear. Of course there is a grain of truth in there but above 6kHz a 'standard' is just a standard so measurements made to that standard are comparable.
As everyone's pinnae and even ear canals differ the squigles in your ears will be at another frequency and amplitude (depending on the headphone and how you put it on as well) than that of a standard. The idea is that the standard is close to an average.
Tons of research has been done on this and deviations of well over 10dB at different frequencies are no exception.
Of course there are many different standard setups, one more anatomically accurate than others (pinna and ear canal) so is why the same headphone measured on different HATS all differ above 6kHz or so. But a standard is a standard and our ears in general are not. Ofcourse some folks will have ears close to that standard and other people don't.

Consider that all measurements have a certain error in it and any 'exact' EQ to squigles, regardless who created them and how, will only have the 'proper' curve on that particular fixture with that particular headphone at a specific (or averaged over seatings) EQ.

For this reason I would suggest NOT to use a single 'accurate' EQ acc to a measurement but look at several measurements and examine the 'average' correction of the biggest 'flaws' seen in different measurements and play with the amount of bass to your liking (seal may be different or taste).

This is why I prefer EQ such as that from Amir (simple EQ just compensating for general errors) than elaborate EQ that, more likely than not, is not accurate for the listener.
Why does that EQ still sound better than no EQ ? Simple... just look at the 'average' EQ. That average difference is what makes it better.

So ignore squigles above 6kHz in plots or at least take them with a heavy dose of salt.
What you see on plots (in the treble range), other than some average tilt, peak or dip, may not be as terrible as it looks in the plots. Also one must realize that HATS all have dips somewhere between 8kHz and 11kHz that can completely obscure peaks that are there but are not shown. Peaks you might well hear and then not EQ.

For IEM measurements the pinna is not used (bypassed) and the ear canal is partly bypassed so that will create some peaks but only from the (shorter) ear canal and not from the pinna which causes nulls (the sharp dips, the squiglies as you will).

 

[MrBlitzpunk]

Ignore squigles above 6kHz or take them with a heavy dose of salt. What you see on plots, other than some average tilt, peak or dip, may not be as terrible as it looks.

So it's more of our anatomical limitations (or in this case, simulated anatomy) then?

I've also heard words like phase cancelation, cup resonance, and free field movements (i don't even know what the latter means) as an explanation for the upper treble squiggles, does this holds merit? What are your thoughts?

 

[solderdude]

Of course, there is also the pinna shape, size, angle of the driver, driver-ear distance, driver size, pads (and how much they reflect/absorb) etc.
As frequencies get higher this will result in cancellations that differ per headphone/pinna/ear canal and used fake pinna, and coupler.

It is a very complex matter and the reason why new HATS (specifically for headphones) still are researched and launched as the next best thing (expensive !!)
The fact that all measurements look different (above 1kHz) is because of all the above mentioned reasons and some more as well. This is not only caused by production spread, test conditions, averaging, etc.

Basically unreliable above 6kHz although some prefer to say above 8kHz and even 11kHz for some specific test fixtures.
No matter how they twist and turn the reasoning and tests... above 6kHz not so accurate and less correlation to how you, I and everyone else may perceive it for whatever reasons they can come up with.

Indicative at best (above 6kHz) but quite accurate below 1kHz (at least it can be) and fairly accurate between 1kHz and 4kHz with somewhat correct pinnae and ear canal and the correct compensation for that pinna and ear canal + target curve.

 

[DVDdoug]

10KHz has a wavelength of about of 1.3 inches. At about 1/2 wavelength the reflected waves start adding or subtracting in complicated ways. At lower frequencies (longer wavelengths) the waves can't really "develop" inside the headphone or inside your ear and the pressure from the driver just gets "nicely" transferred to your eardrum.

with in-ears the distances are shorter so things shouldn't get strange until you get to higher frequencies, perhaps above the audio range.

 

[MrBlitzpunk]

Of course, there is also the pinna shape, size, angle of the driver, driver-ear distance, driver size, pads (and how much they reflect/absorb) etc.

Now you got me wondering how a headphones measurements would look like without the simulated ear

 

[solderdude]

Just look at measurements on my website . They are all without pinna and ear-canal (exactly for the mentioned reasons) but are inaccurate between 1kHz and 5kHz. How much the deviation is depends on the angle of the driver, driver -ear distance and driver diameter (and modal distributions).
Note my test fixture incl. electronics and ADC are < $ 100.- so take that (and the fact it was out of hobby logging measurements) into consideration before taking it every bit as serious as all other 'headphone measurement attempts' out there.

 

[MrBlitzpunk]

Just look at measurements on my website .

Thank you! That's a very good read actually, thorough and easy to understand. Learn a couple of things too, i just now realized that the waterfall plot is meant to measure resonance and decay, yeah?

 

[solderdude]

Waterfall plot can show where resonances are and how long they last.
In general the decay of a note from an instrument is longer than that of a headphone so in most cases it does not say much other than there is a resonance at certain frequencies.
Hifiman planars can have tons of resonances that are long lasting yet isn't as audible as some might think.

 

[Deleted member 16543]

Waterfall plot can show where resonances are and how long they last.
In general the decay of a note from an instrument is longer than that of a headphone so in most cases it does not say much other than there is a resonance at certain frequencies.
Hifiman planars can have tons of resonances that are long lasting yet isn't as audible as some might think.

Compare this statement to the "night and day" differences self-proclaimed golden ears claim to hear when the subject is higher sample rate or bit depth, and you have a pretty accurate snapshot of the sea of BS one has to navigate before landing on some truth, when researching audio online.
Then get a subscription to some audio magazine that peddles $multi-k power supplies and cables, and you realize you need to adjust and multiply that initial estimate by 1 million.
But it's sooooo entertaining!