How do You Evaluate Treble Measurements?

We'll probably all agree that treble measurements are noisy and difficult to interpret. It's mostly blurring your eyes to get the gist of the general trends. Ah well, please select which best describes what you see and are looking for in treble response measurements.

Feel free to elaborate on how useful treble headphone measurements are to you in the comments.

After submitting your vote, you're done. Thank you for helping out!

How do You Evaluate Treble Measurements?
Can we get a holla for a flat line! Okay, maybe a little roll-off.
23% (28 votes)
Heaven forbid an upward spike between 4-8kHz; after that just a little roll-off.
35% (43 votes)
I like the evermore-descending response of the Harman target, all the way up.
27% (34 votes)
I really can't tell what's going on up there.
9% (11 votes)
Don't believe anything you see after 4kHz!
6% (8 votes)
Total votes: 124

COMMENTS
ADU's picture

Already mentioned this in the introductory post to the polls. But imo the Harman curve falls short in the treble because it does a poor job of modeling the resonances in the head & torso simulator at 2.7-3.5 kHz, 9-10 kHz, and 15-16 kHz.

These HATS resonances show up fairly often on raw plots of neutral and close-to-neutral headphones, esp. when you begin averaging the curves together to get a better overview of what's going on. A few examples...

http://www.innerfidelity.com/images/FocalSpiritProfessional.pdf
http://www.innerfidelity.com/images/NADVISOHP50.pdf
http://www.innerfidelity.com/images/OppoPM3SampleC.pdf
http://www.innerfidelity.com/images/SennheiserHD580wHD600headband.pdf

I voted for the Harman treble response though, because I think its overall shape is still closest to neutral and my own preferences of the poll options.

ADU's picture

I meant "falls short" figuratively btw. Imo, the Harman curve probably over-estimates the ideal response in some spots in the treble, esp. around 5 to 7.5 kHz, where there's usually a bit of a dip between resonances.

ADU's picture

I think Tyll meant "_treble_ response measurements" rather than "midrange..." in the description above btw.

Tyll Hertsens's picture
Fxt, thanks!
Aufdemaury deus ex machina's picture

Not much point in picking up a ruler for a headphone in a quest to find a headphones regarding FR response, since there's a large amount of equalization in recordings anyways, and the very monitors/headphones that one would use to reference a track in a recording setting varies too,
so in truth, even though you have a ruler flat FR response, you will most likely not hear what was intended anyway. Of course these recordings are also compensated to fit and best a broader consumer market so it won't sound half bad on supermarket buds.

In my experience never once have i've seen a single person IRL use the hd800's for anything regarding the production of music, despite it being a great tool, it becomes moot when everyones listening to your music on different mediums. Though i have seen hd600/650's, Audeze's, Akgs's Beyer's but other than that nothing amongst the above an beyond the 300$ tag.

It's the same exact story with digital/vs analog

some studious used mixed digital and analog techniques such as time correction on digital means, and overdubs on tape etc.

that to is a extraordinarily moot topic as well.

moopster's picture

Check out soundliaison.com, they are audiophiles and list the equipment used in their recordings, including the HD 800.

zobel's picture

Is bass 10 Hz to 250 Hz?
Is midrange 250 Hz to 2500 Hz?
Is treble 2500 Hz to 20 kHz?

or what?

Tyll Hertsens's picture
I've seen it described a number of ways. I use John Atkinson's numbers, which are, as I recall: bass -200Hz; kids 200-1200Hz; treble 1200Hz-.
zobel's picture

The reason I chose those points came from using a test CD from JA that divided them up that way.

That was handy for me because the test tones breaking into treble at 2.5kHz is where most tweeters start, at the lowest.

My kids were in the range you mentioned above, until they reached puberty :-)

If anything, from what I've been raking up here online, the 200Hz bass/mid point is very popular, but the mid/treble point is almost always considered higher, at least 2.5kHz.

It isn't a huge deal, but we would benefit by agreeing on;

1) sub-bass...
2) low bass....
3) mid bass....
4) upperbass....
5) low mids...
6) midrange...
7) upper mids...
8) low treble...
9) mid treble..
10) upper treble..

You should make that call Tyll. Maybe it would be good to post this definition in the measurements page in Resources? That way we will all be on the same page and have a common vocabulary.

Happy Spring!
zobel

wink's picture

There's nothing worse than a strident, grating and banshee emulating sound.
In real life, the highs of a piano or flute can hurt you right down to the roots of your teeth.
If that's what is supposed to be reproduced, suck it up.
If not, get rid of the Grado and get a noice HD600.

Mauro's picture

There is still some difference between my NAD HP50 and my KEF LS50 in the treble and is not just about decibels. The latter is my favourite, but it is difficult to elaborate on this without a flat line to EQ. I have tried with the Harman software for PC but without luck..I guess that researchers should still figure out what's the desired treble response and manufacturers how to realize it (and remove any unwanted wiggle...)

castleofargh's picture

it depends on a lot of things. on your graphs I would still care about what's happening at 10 or 12khz, but on the stuff I measure at home with my "el cheapo" devices and more than 20db boost for calibration in the upper trebles to pretend like the graph is a proper RAW reading, I trust maybe up to 8khz, and only frequency response(+/-3db I'd say?). above 8khz is only really giving me an idea of where the usual massive roll off is starting. kind of informative for further EQing but that's about it.

also I would make a difference between headphones and IEMs. changing tips or insertion depth on IEM can change so many things that I rely on headphone measurements a lot more for trebles.

Brad331's picture

If not, wouldn't there be a discrepancy?

ADU's picture

And that's precisely the point I was trying to make in my first post above.

Every head and ear is different, and will have its own unique set of resonances and measured response to sound. As long as the headphones are all measured on the _same_ head and ears though (as in the IF database), then you can still make useful comparisons from the data, because you're comparing apples to apples.

Comparing the in-ear measurements from two _different_ heads though is like comparing apples to oranges. You may still be able to get some general idea of the differences between the headphones. But many of the crucial details (especially in the treble) will be lost in the differences between the two "systems" used for measurement.

To put it simply, the problem with the current Harman curve, imo, is that the in-ear measurements were taken with "a different head" than the one used by Inner Fidelity. The head and torso simulator that Harman used to develop their current target headphone response curve does not seem to exhibit the same characteristic set of resonances in the treble as the simulator used by Inner Fidelity, or for that matter by Golden Ears. That makes it a poor model for the ideal Inner Fidelity treble response imo, because its like comparing apples to oranges, rather than apples to apples.

The head and torso simulators used by Harman and Inner Fidelity are designed to conform to certain standards for an average sized head and ears. But those standards apparently aren't stringent enough to yield comparable results in the treble in the two systems. Averaging together the raw FR curves of several different good-sounding headphones in the IF database will imo produce a much clearer picture of the characteristic resonances in the IF head and torso simulator.

The raw FR plots at Golden Ears are also slightly different than the Inner Fidelity plots btw. The main peak in the lower treble usually occurs at a slightly lower frequency in the Golden Ears plots than it does in most Inner Fidelity plots, for example. The GE plots are generally alot closer in overall shape to the IF plots than the Harman plots are though. And the GE plots have virtually the same set of peaks and valleys in the treble, on average, as the IF plots.

Bottom line for me is that the Harman curve, in its current form, is not a very reliable or accurate model for the correct treble response on either the Inner Fidelity or Golden Ears FR graphs.

Bruch's picture

In the 2nd option do you mean that we are voting for or against an upward spike between 4-8kHz? I think that it could be read both ways. I voted on the basis that it meant against.

ADU's picture

Pretty sure it's a vote against the excessive spikes in that range.

Bob Katz's picture

I need to see a 1/6 octave filtered curve of variation from the Harman. Then the measurement will make a lot more sense to me.

ScaryFatKidGT's picture

I actually like high treble and air and presence, I don't find it harsh like the 4-8khz range

X