[tinyman392]

Yup, exactly that.

What defines a preferred sound in headphones? I think the video is saying the preferred sound is more akin to the sound of flat speakers?

 
LOL, I was updating my post while you ninja quoted it  If I'm not mistaken, Tyll did report that users preferred a non-neutral speaker signature as well.  He described the signature as a linear line with a 10 dB bump in the bass and a 0 dB gain in the treble.  
 
Essentially, they also retested flat speakers vs non-flat.  The surprising thing is that the signatures of the headphones and speakers still didn't match up  The results are interesting, but should also be noted, there was a small sample size.  
 
It's been a while since I looked at the stuff regarding the paper, please correct me if I'm wrong.  

 

[castleofargh]

  The Olive Welti experiments were designed to find the most preferred sound, with respect to music, regardless of sound signature type (e.g., any sound signature).  They were not designed to find the most neutral sound, it shouldn't be read that way (it wasn't its focus or intent).  They wanted to find out what people's preferred sound signature was, and they determined that it was non-neutral.  I should also note that in terms of speaker response, the preferred also was non-neutral, rather, they also preferred a warm tilt.  With headphones, they preferred slightly more bass, specifically sub-bass.  
 
Now we have to figure out why they prefer that sound over neutral (which Etymotics are not DF neutral still, they tone down the treble by a few dB to compensate for compression which has gone way up since then).  IMO, I feel that this preference towards a warmer sound is due to compression.  But, that's just speculation.  

in fact from what they did years ago on speakers they did find that people tended to prefer flat speakers (real flat). what wasn't flat is the response we get where we are sited in the room (room reflection+ high frequencies getting attenuated in the air faster than low frequencies over a distance). so in a way the slightly warmer FR is a perfectly natural phenomenon in real life listening, be it from speakers or a real band, and the farther away we are from a band the warmer they will sound.
 
and what they tested for headphone too was as you say, what people preferred (they did several trials in several countries, not just the 10 employees or something at the beginning). but they started with the assumption that people would like to have in headphones the sound they prefered from flat speakers, so with the tilt in FR from the start as this time air couldn't do the trick by itself in headphones. and they seemed from the results to be right about that. so the "prefered" headphones end up with the warm tilt to simulate the change that would occur if there was more distance between the drivers and the ears.
 to me etymotic worked to get neutral sound without a care in the world if it was natural or not(and that's why they are so cool as monitoring IEMs), while olive worked to get the sound from neutral speakers in that one cool room at that one distance that is their "ideal" listening room.
so if the band was playing on my lap and the singer singing into my ear, then I would think that etymotic got it right. but if the band is supposed to be on stage and me 30 or 40meters away, then Olive certainly has the more realistic idea and I would expect the sound to have maybe 2 or 3DB tilt from 20hz to 20khz.
 
but yeah I should be careful when I write neutral or flat, as it can mean so many things.

 

[MrMateoHead]

 
I think I originally stole this from innerfidelity.

 
the black one is what people preferred/found neutral for headphones. so if your raw measurement of headphone (the gray lines on tyll's graphs) looks like the black curve here, harman claims from the study that it is what most people will find neutral and prefer.  the main conclusion of the studies at harman showed pretty much that all people of all ages of all races did tend to favor neutral sounding gears, as in with the same FR that from flat speakers in an ideal room once the sound has travelled toward us. so not really neutral anymore as trebles get attenuated with distance.
 
compared to the usual compensation curves, this one does ask for a bass boost, and does let the trebles go down as they would naturally in the air when coming from speakers. 
to make it simple, you take the graphs from almost everywhere, and when they show something like a slow regular down slope from bass to trebles, chances are it would be close to flat on harman's target curve.
think LCD2 as closer to flat for them than ER4 or HD800. something I clearly tend to agree with ^_^.
 
http://seanolive.blogspot.fr/2014/01/the-perception-and-measurement-of.html   this video is a "short" version of the 2 main papers done on the subject.

Just a quibble -
 
Harmon's research, while generating the new curve, did NOT conclude that "all people of all ages / races" expressed a preference for that particular curve. The population administered the listening test was far, far smaller than that, and I don't recall being representative of the general population (as defined by average age, sex, or race, for example, or trained / untrained listening skills). While their approach, and conclusions should be repeatable, that study was not the end-all be-all of sound perception studies.

 

[miceblue]

 


I think I originally stole this from innerfidelity.



the black one is what people preferred/found neutral for headphones. so if your raw measurement of headphone (the gray lines on tyll's graphs) looks like the black curve here, harman claims from the study that it is what most people will find neutral and prefer.  the main conclusion of the studies at harman showed pretty much that all people of all ages of all races did tend to favor neutral sounding gears, as in with the same FR that from flat speakers in an ideal room once the sound has travelled toward us. so not really neutral anymore as trebles get attenuated with distance.

compared to the usual compensation curves, this one does ask for a bass boost, and does let the trebles go down as they would naturally in the air when coming from speakers. 
to make it simple, you take the graphs from almost everywhere, and when they show something like a slow regular down slope from bass to trebles, chances are it would be close to flat on harman's target curve.
think LCD2 as closer to flat for them than ER4 or HD800. something I clearly tend to agree with ^_^.

http://seanolive.blogspot.fr/2014/01/the-perception-and-measurement-of.html   this video is a "short" version of the 2 main papers done on the subject.

Just a quibble -

Harmon's research, while generating the new curve, did NOT conclude that "all people of all ages / races" expressed a preference for that particular curve. The population administered the listening test was far, far smaller than that, and I don't recall being representative of the general population (as defined by average age, sex, or race, for example, or trained / untrained listening skills). While their approach, and conclusions should be repeatable, that study was not the end-all be-all of sound perception studies.

Did you take a look at the video overview?

 

[castleofargh]

Just a quibble -  
Harmon's research, while generating the new curve, did NOT conclude that "all people of all ages / races" expressed a preference for that particular curve. The population administered the listening test was far, far smaller than that, and I don't recall being representative of the general population (as defined by average age, sex, or race, for example, or trained / untrained listening skills). While their approach, and conclusions should be repeatable, that study was not the end-all be-all of sound perception studies.

 
 
 
the trial about races/countries was one test on it's own. age was another one. and did involve hundreds of people. the conclusions were that on average people tended to like the same thing.  not like it's actually a surprise. the "young people love bass" myth has brothers in every technological fields, kodac made different batches of the same camera films for different countries for no real reason except urban legend and force of habits (Germany more green, USA more red...). surprise we're all humans!!!!
then other tests showed that the people tested liked the sound of neutral speakers in a room the best.
then another test showed that trying to use the perceived signature of neutral speakers in a room, on headphones, made the test subject to mostly favor that signature over diffuse field and free field calibration and even over the natural headphones signature.
other tests made people swap headphones, another simulated different headphone signatures in on particular model, another test just gave an EQ to people and let them play with it. I think that over the last year(maybe a little more), harman has been at it and passed the statistically significant level.
all those tests have a common idea and all end up being conclusive for that common idea.
so if we tend to all like pretty much the same thing, and if test subjects preferred that one signature to the usual compensation curve used, or other already existing headphones, am I wrong in saying that the conclusion of the tests is pretty much that people of all age and race tend to favor that curve?
 I wasn't trying to sell anything, I just made some simplified explanation for zorrofox who didn't hear about it. there is I think 2 topics about, at least one of the trials on headfi and a lot of PDFs available on Sean Olive's blog, and on AES.
 
 
anyway I think I've unintentionally made a long enough out of topic, sorry guys let's odac/o2.

 

[MrMateoHead]

Did you take a look at the video overview?

Unless I missed something - the summary says they used 11 trained listeners in the first test. They then also used approximately 200 persons trialed with "virtual" headphones to test preferences of persons from different nationalities / ages and so on. My quibble is that one shouldn't conclude that Harmon decoded "global" preference for headphone frequency response - that conclusion is too broad given the limitations of the study and its focus, and anyway an $8 billion dollar industry could be composed by well over a million people. Their effort to actually do R&D on headphone response curves is great, but even Harmon admits that there are other, measurable factors that weren't being considered (like headphone distortion). Obviously, if "Headphone A" had a perfect, subjectively neutral response curve, but couldn't playback at 70dB without major distortion (while Headphone B could), a sample of the population would probably hate it (the hard of hearing aka older persons, for example). Science is great, but more work is needed, obviously. Furthermore, knowing what might be preferred doesn't mean that a company will actually MAKE the thing - given that target response would be just one factor in product development process (which has to be delivered at a competitive cost).
 
I don't dispute their findings - but I just wouldn't declare them 'global' either, and that is what the poster seemed to be implying.

 

[castleofargh]

 

Did you take a look at the video overview?

Unless I missed something - the summary says they used 11 trained listeners in the first test. They then also used approximately 200 persons trialed with "virtual" headphones to test preferences of persons from different nationalities / ages and so on. My quibble is that one shouldn't conclude that Harmon decoded "global" preference for headphone frequency response - that conclusion is too broad given the limitations of the study and its focus, and anyway an $8 billion dollar industry could be composed by well over a million people. Their effort to actually do R&D on headphone response curves is great, but even Harmon admits that there are other, measurable factors that weren't being considered (like headphone distortion). Obviously, if "Headphone A" had a perfect, subjectively neutral response curve, but couldn't playback at 70dB without major distortion (while Headphone B could), a sample of the population would probably hate it (the hard of hearing aka older persons, for example). Science is great, but more work is needed, obviously. Furthermore, knowing what might be preferred doesn't mean that a company will actually MAKE the thing - given that target response would be just one factor in product development process (which has to be delivered at a competitive cost).
 
I don't dispute their findings - but I just wouldn't declare them 'global' either, and that is what the poster seemed to be implying.

but you're just being picky on words. my sentence was "the main conclusion of the studies at harman showed pretty much that all people of all ages of all races did tend to favor neutral sounding gears, as in with the same FR that from flat speakers in an ideal room once the sound has traveled toward us." again I was simplifying to make it short in English when I'm a surrendering frog by birth.
my sentence uses "pretty much all" instead of all, "tend to favor" instead of "blindly followed the global rule of humanity", and doesn't even talk about that curve. only about a general idea. it's a study, we're humans and yes we're all free to like whatever we want. it doesn't say that more bass can't be fun, or that more trebles can be nice on classical... I didn't think I was a dictator when I wrote that sentence.
I read "X product is the bestest" all day long on headfi even when it's a placebo product, I didn't think my sentence would be such a problem. diffuse field or slightly modified diffuse field is what is used to show 99% of headphone and IEM measurements nowadays, apply harman's curve (that isn't night and day different) and tell me it doesn't seem more accurate about what flat sounds like. to me(and a great deal of people I talked to) diffuse field compensation makes bright headphones look flat.

 

[Dark_wizzie]

What is the Harman Curve and how is it derived? I recall that the human ear is more sensitive to some frequencies than others, so if we want a perfectly neutral sounding headphone for example, would the FR curve have be closer to an equal-loudness-contour than a flat line on a FR graph? 

 

[castleofargh]

  What is the Harman Curve and how is it derived? I recall that the human ear is more sensitive to some frequencies than others, so if we want a perfectly neutral sounding headphone for example, would the FR curve have be closer to an equal-loudness-contour than a flat line on a FR graph? 

I think we should stop hijacking this topic.
the curve is about the first graph I have posted.
at the moment what I'm advocating for, is for us to use harman compensation curve for headphone graphs instead of basic diffuse field as it's done right now in most cases. but it's just a way to show the frequency response, the measurements are still the same.
flat raw measurements are the real deal flat as in "set by science" flat. then we try to find compensation curves for our faulty ears ^_^(that would be related to equal loudness contour and will slightly deviate from one person to another). then we need to compensate depending on the source of sound.
the problem has been solved years ago for speakers(by harman), but I think it isn't yet solved for headphones, and it isn't for IEMs. and that curve to me shows something closer to what I actually hear compared to diffuse field compensation.

 

[adydula]

Please take this topic to its own thread and lets get back to the O2/ODAC.
 
Thanks
Alex

 

[faisal2003456]

So, I'm on the JDS labs site and I'm wondering.... Which gain set-up for the HD600? I figure the lower the gain, the less distortion I'll get. Oh also, I listen to music at relatively high volumes I guess.
 
I was looking at some Audio-GD and Schiit products, but I'm a college student who moves between home and the dorms A LOT. I need a transportable single-unit device and the O2+Odac fits my size requirements. I saw some Nuforce products for cheap , like the Icon DAC and Icon HD (HDP without the extra inputs); but man.... Nuforce's reputation is trashed. I don't see any reviews of their stuff anymore. I'd rather not take a gamble.

 

[Solrighal]

I use gains of 1x/3x with my HD 650 and I only rarely feel a need to use the higher setting.1x is generally more than enough.

 

[jring]

  So, I'm on the JDS labs site and I'm wondering.... Which gain set-up for the HD600? I figure the lower the gain, the less distortion I'll get. Oh also, I listen to music at relatively high volumes I guess.
 
I was looking at some Audio-GD and Schiit products, but I'm a college student who moves between home and the dorms A LOT. I need a transportable single-unit device and the O2+Odac fits my size requirements. I saw some Nuforce products for cheap , like the Icon DAC and Icon HD (HDP without the extra inputs); but man.... Nuforce's reputation is trashed. I don't see any reviews of their stuff anymore. I'd rather not take a gamble.

 
Hi,
 
if you plan to use ODAC and O2 always in combination and also on the go, 1x and 2.5x are your best bet. 2.5x is the highest gain which works with a high voltage source like ODAC (2V redbook) on batteries. If you are using it powered and with ODAC, 3.5 is the limit. If you use higher gain than 2.5 on batt or 3.5 powered with ODAC it will distort regardless of the O2 volume setting.
 
All the higher gain settings are for low output sources only.
 
On the other hand you can get the default config of 2.5 and 6.5 for maximum flexibilty and see if 2.5 is ok for you. If you see you need a finer volume control with high sensitivity phones, you could clip the high gain resistors to give you 1x for the high setting (slight cosmetic problem here).
 
I for my part have not been able to get anywhere close to 12 o'clock with ODAC on 2.5x regardless of cans and my sextetts are not really easy to drive...
 
Joachim

 

[faisal2003456]

Thank you for the replies. I couldn't find a battery powered O2+ODAC (single unit) that was pre-assembled. Where can I find one? I'd prefer a battery, but wall-powered is fine. The brick is pretty small. I was thinking about either:

The 1.0X and 2.5x or
The 1.0X and 3.5X
 
So I was never considering the 6.5x (which is ridiculous).

If the 2.5x can get the HD600/650 to ear-splitting volumes with no distortion then I'll just get that.
 
Also, I was thinking about get the model with the RCA outputs on the back to run into a speaker-amp (if I ever get one). Is there some kind of auto-switching in place?

Cheers!

 

[jring]

  Thank you for the replies. I couldn't find a battery powered O2+ODAC (single unit) that was pre-assembled. Where can I find one? I'd prefer a battery, but wall-powered is fine. The brick is pretty small. I was thinking about either:
  Also, I was thinking about get the model with the RCA outputs on the back to run into a speaker-amp (if I ever get one). Is there some kind of auto-switching in place?

 
The normal O2 ODAC combos won't have batteries since the ODAC goes where the batteries are. I have see some hack jobs of putting the ODAC under the O2 board...
 
RCA plugs usually don't have a switch unlike the 1/8" plug...
 
Joachim