IEMs close to Harman curve?

[daftcombo]

Listeners in the test were asked to say what they preferred. Listeners were from all walks of life and different ages. The summary reflects how all of them judged a headphone subjectively, just as well as you and I would if we had participated. The difference between trained and untrained listeners was not significant in what they prefered.

I understand but that sounds like we consider objectivity to be a shared subjectivity (better, an averaged subjectivity).
Which could lead to state earth is flat even now

Do you know how much people were involved, standard deviation.. ?

 

[amirm]

Do you happen to know at which average SPL these tests were done ?

Yes, they used the ITU recommendation for 2-channel testing which is 82 dB SPL:

 

[amirm]

Do you know how much people were involved, standard deviation.. ?

Sure:

Standard deviation is shown as error bars on the preference ratings:

As you see, they are quite small relative to overall score and would not change the outcome/rankings. There is a ton of statistical analysis in the paper. Almost all the outcomes reached and exceeded 95% confidence (p<0.05).

 

[jazzendapus]

I understand but that sounds like we consider objectivity to be a shared subjectivity (better, an averaged subjectivity).
Which could lead to state earth is flat even now

Do you know how much people were involved, standard deviation.. ?

How do you like that phrasing: it's an objective fact (well, almost , I guess) that people prefer a particular FR of headphones. Pick a random person off the street and there's a very high chance he'll prefer headphones with FR that follows the Harman curve.

 

[daftcombo]

Thank you. How do you read the graph?

 

[amirm]

Thank you. How do you read the graph?

You mean the blue bar graph? If so, that is the composite score for all the listeners for each individual headphone. Higher is better.

Low-anchor is a crappy headphone and is there to make sure the listener can find that one. And indicates the test is working.

Target is the simulated response in a virtual headphone (I think). It garnered the highest preference. The rest are actual headphones and how they were rated. Headphone 1 (HP1) did the best. The worst, HP6, did even worse than the crappy headphone!

 

[hetzer]

That's basically what Harman is all about, averaged preference of listeners.
With there being a small Delta between the subjects this seems like a good way to test.
Do you happen to know at which average SPL these tests were done ?
Is it known how big of a delta there is between experienced listeners and casual non trained ones ?

The loudness is matched at ITU 1770.2 to 78dB

 

[daftcombo]

How do you like that phrasing: it's an objective fact (well, almost , I guess) that people prefer a particular FR of headphones. Pick a random person off the street and there's a very high chance he'll prefer headphones with FR that follows the Harman curve.

I agree but:
The average Joe, the random person in the street used to MTV might like a lot of bass, which could not correspond to a flat FR in an anechoic room.

 

[amirm]

I agree but:
A random person in the street might like a lot of bass, which would not correspobd to a flat FR in an anechoic room.

The research as with speakers, shows that the random person is the same as an audiophile, or an engineer.

 

[hetzer]

The loudness is matched at ITU 1770.2 to 78dB

I saw amirm's post after i wrote this post. In 'The Relationship between Perception and Measurement of Headphone Sound Quality', It is matched at ITU 1770.2 to 78dB B-weighted. It seems like the loudness is not the same between the tests.

 

[daftcombo]

You mean the blue bar graph? If so, that is the composite score for all the listeners for each individual headphone. Higher is better.

Low-anchor is a crappy headphone and is there to make sure the listener can find that one. And indicates the test is working.

Target is the simulated response in a virtual headphone (I think). It garnered the highest preference. The rest are actual headphones and how they were rated. Headphone 1 (HP1) did the best. The worst, HP6, did even worse than the crappy headphone!

Thank you, it is clearer as you explained the abbeeviations but not completely clear.
I don't manage to put in words what I see.
For instance, should we understand that some people prefer the HP6 for some songs? (which would be perfectly compatible with stat results)

 

[hetzer]

'The most prefered' does not mean that it is 'Best for everyone and every circumstance'. It is just same as politics. A president who is elected with 50% vote does not satisfy every citizen during his(or her) term of office.

 

[daftcombo]

The research as with speakers, shows that the random person is the same as an audiophile, or an engineer.

Ok then I guess the IEM curve could be explained by the inner shape if the ear and, once this latter modelized and thrown into the equation somehow, make us find a rather flat target again, and let us improve the sound again with reaching to a flat one.

 

[amirm]

For instance, should we understand that some people prefer the HP6 for some songs? (which would be perfectly compatible with stat results)

Three different tracks were used that were shown in prior research to be most revealing in such tests. The ratings persisted across them:

By "program" they mean music track used for testing.

 

[amirm]

Ok then I guess the IEM curve could be explained by the inner shape if the ear and, once this latter modelized and thrown into the equation somehow, make us find a rather flat target again, and let us improve the sound again with reaching to a flat one.

Yes, except that with IEMs, the research says we want more than "flat" when it comes to bass.

One issue with the test which is acknowledged is that the levels were not adjusted down when the bass was increased. This may have tilted preference due to loudness more than just the bass.

 

[solderdude]

I saw amirm's post after i wrote this post. In 'The Relationship between Perception and Measurement of Headphone Sound Quality', It is matched at ITU 1770.2 to 78dB B-weighted. It seems like the loudness is not the same between the tests.

one is c weigthed and the other one b weighted.
As most of the energy is in the lows it is easy to get different numbers yet it could play equally loud.

 

[daftcombo]

Yes, except that with IEMs, the research says we want more than "flat" when it comes to bass.

One issue with the test which is acknowledged is that the levels were not adjusted down when the bass was increased. This may have tilted preference due to loudness more than just the bass.

This a crucial information.
Also, I've read that the 3kHz zone that is boosted here was critical for the perception of loudness.

 

[daftcombo]

Yes, except that with IEMs, the research says we want more than "flat" when it comes to bass.

Bass in IEM heavily depends on how they are settled, if they fill the earhole or let air pass.
It must be really difficult to measure.

 

[solderdude]

Bass in IEM heavily depends on how they are settled, if they fill the earhole or let air pass.
It must be really difficult to measure.

probably easier to measure consistently (always the exact same diameter/shape and insertion depth) than getting the same seal/bass response with variances between person to person as well as insertion depth.

 

[Soniclife]

probably easier to measure consistently (alwasys the exact same diameter/shape and insertion depth) than getting the same seal/bass response with variances between person to person as well as insertion depth.

I would hope they also test to see if the isolation is consistent, which is how I think most humans can tell if they are seated correctly, if the outside world isn't attenuated it's not fitting right.