[bigshot]

Haven't had a chance to read this yet, but someone suggested it to me on the Head-Fi Facebook page and it looks interesting. This is different from what I would expect.

The Journal of the Acoustical Society of America: “No Correlation Between Headphone Frequency Response and Price.”
https://asa.scitation.org/doi/full/10.1121/1.4984044

Another interesting study where they tried to EQ midrange cans to sound like a variety of other brands and models
https://www.aes.org/e-lib/browse.cfm?elib=16874

 

[bfreedma]

Haven't had a chance to read this yet, but someone suggested it to me on the Head-Fi Facebook page and it looks interesting. This is different from what I would expect.

The Journal of the Acoustical Society of America: “No Correlation Between Headphone Frequency Response and Price.”
https://asa.scitation.org/doi/full/10.1121/1.4984044

Another interesting study where they tried to EQ midrange cans to sound like a variety of other brands and models
https://www.aes.org/e-lib/browse.cfm?elib=16874

Interesting, but the full studies are paywalled. Since you’ve got access, can you provide a brief synopsis of the testing and actual results?

 

[bigshot]

I don’t have access. Maybe someone would like to take one for the team?

 

[castleofargh]

Olive's blog on the second paper: http://seanolive.blogspot.com/2016/04/a-virtual-headphone-listening-test.html
As the test actually uses the same headphones with convolution to mimic the others, it's easy to argue that there is more to it when listening to the actual headphones and choosing our preferrence.

And while we're at it, similar question about preference and price relation with some IEMs: http://seanolive.blogspot.com/2017/02/twirt-337-predicting-headphone-sound_17.html

 

[bfreedma]

I don’t have access. Maybe someone would like to take one for the team?

We‘re all saving for audiophile network switches and quantum stickers, so no extra cash...

When I have some more free time to dig in, I plan on getting an AES membership.

 

[bfreedma]

Olive's blog on the second paper: http://seanolive.blogspot.com/2016/04/a-virtual-headphone-listening-test.html
As the test actually uses the same headphones with convolution to mimic the others, it's easy to argue that there is more to it when listening to the actual headphones and choosing our preferrence.

And while we're at it, similar question about preference and price relation with some IEMs: http://seanolive.blogspot.com/2017/02/twirt-337-predicting-headphone-sound_17.html

Hmmm. Not sure that using a single headphone with convolution gets me there.

Though I do understand the challenge. Got a chuckle (as an Audeze owner) when they mentioned that the weight of the LCD2 gave it away and made blind testing of the actual units impossible. Not an easy problem to solve.

 

[bigshot]

I don't know about virtual listening tests, but it sure shows that they seem to accept that midrange cans can be processed to sound like high end ones. Try saying that outside of Sound Science on Head-Fi!

When I was corresponding with the designer of the Oppo PM-1, he told me that the major difference between expensive headphones and much less expensive ones was consistency between samples. He said that the PM-1s were all tested as they came off the line, and if they weren't with +/-1dB of the target curve, they were rejected. He said that the acceptable deviance for most manufacturers was usually 3dB, and in some cases as much as 5dB. Oppo was cherry picking off the line and taking the ones that didn't measure up as a loss. (Well probably not entirely a loss- they came out shortly with a lower priced PM-2 and PM-3 that I suspect were made from the guts of PM-1s that didn't make the cut.) The study didn't check sample variation.

 

[ADUHF]

Interesting study. Thanks for posting this.

 

[ADUHF]

I wish they had used diffuse field in addition to, or instead of raw measurements for this study.

 

[ADUHF]

Haven't had a chance to read this yet, but someone suggested it to me on the Head-Fi Facebook page and it looks interesting. This is different from what I would expect.

The Journal of the Acoustical Society of America: “No Correlation Between Headphone Frequency Response and Price.”
https://asa.scitation.org/doi/full/10.1121/1.4984044

This study only looks at one aspect of sound quality btw. Namely, the frequency response. Which is certainly an important factor, but not the only I'd consider when buying a new headphone.

 

[bigshot]

It's been my experience that frequency response is by far the most important consideration with transducers. Everything else is relatively insignificant in comparison. The study itself mentions this.

 

[castleofargh]

This study only looks at one aspect of sound quality btw. Namely, the frequency response. Which is certainly an important factor, but not the only I'd consider when buying a new headphone.

Other factors in sound that could influence preference have been researched, but I don't know of anybody trying to correlate that with headphone prices. There are several possible reasons for a product to be expensive. From the get go, only the brainwashed consumers that we all are, would assume a direct correlation between price and sound preference. Some headphones might be expensive because of the brand. Some might be done in small quantities. Some might use parts that are expensive to produce. Some might lead in specific aspects of fidelity but have a frequency response that doesn't necessarily please a majority of people.


About distortions and preferences, I know of 2 papers involving Steve Temme:
"A New Method for Measuring Distortion using a Multitone Stimulus and Non-Coherence"
by Steve Temme and Pascal Brunet
It's interesting because they suggest that there are more relevant measurements than THD. And also the method allows music as test signal. Which is an obvious advantage when trying to correlate measurements with subjective impressions of sound.
But I have no idea how to run such a measurement, and you won't see it used very often. Maybe it's not that relevant compared to FR, or perhaps it's not simple to do for the average Joe? I sure got overwhelmed by the math.

The second paper is:
"The Correlation Between Distortion Audibility and Listener Preference in Headphones"
Steve Temme, Sean E. Olive, Steve Tatarunis, Todd Welti, and Elisabeth McMullin

Here is a video on the second paper. A bunch of interesting stuff in that video IMO, if you can survive the very annoying sound quality.

 

[HiFiHawaii808]

Haven't had a chance to read this yet, but someone suggested it to me on the Head-Fi Facebook page and it looks interesting. This is different from what I would expect.

The Journal of the Acoustical Society of America: “No Correlation Between Headphone Frequency Response and Price.”
https://asa.scitation.org/doi/full/10.1121/1.4984044

Another interesting study where they tried to EQ midrange cans to sound like a variety of other brands and models
https://www.aes.org/e-lib/browse.cfm?elib=16874

Their findings are not surprising to me at all. The various product manufacturers are not trying to tune their headphones to a target curve unless it is to its own internal target curve. And, the mean deviation from the curve is not as important as fine details around certain frequency ranges that give the headphones certain attributes such a clarity. If you want to create a clean bass shelf, for instance, you want a dip the FR curve between the bass and the mids so that bass doens't bleed into the mids. If you want more airiness to your treble, you boost somewhere in the 8000-10000 hz range. If you are listening to classical music vs Hip Hop, the preference curve is different by those genres. So, it's not useful to correlate a target curve to price.

In addition, there are attributes of sound that don't show up so readily in the frequency response like Sound stage size, resolution, layering, timbre, etc and these all play a part in terms of how we perceive sound.

Frequency response is probably the number 1 variable to determining quality of sound, but not the only one.

What is useful is tuning several headphones to the same target curve. I've done this. The headphones sound similar. The differences are in the other attributes of sound that are also important in determine the quality of a headphone.

 

[ADUHF]

The second paper is:
"The Correlation Between Distortion Audibility and Listener Preference in Headphones"
Steve Temme, Sean E. Olive, Steve Tatarunis, Todd Welti, and Elisabeth McMullin

Here is a video on the second paper. A bunch of interesting stuff in that video IMO, if you can survive the very annoying sound quality.

Interesting study. Thanks for sharing this, castleofargh.

 

[bigshot]

In addition, there are attributes of sound that don't show up so readily in the frequency response like Sound stage size, resolution, layering, timbre, etc and these all play a part in terms of how we perceive sound.

All of those things are covered by frequency, amplitude and timing. And any deviation from fidelity would be distortion. Frequency and amplitude are the big hurdles for headphones. Timing and distortion are bigger issues for speakers.

What is useful is tuning several headphones to the same target curve. I've done this. The headphones sound similar. The differences are in the other attributes of sound that are also important in determine the quality of a headphone.

Since it's pretty much impossible to do a blind test of headphones, it's hard to test for that or quantify it. But I would bet that a listener's ears would accommodate to those things pretty easily, and when listening to commercially recorded music, it wouldn't be very noticeable.