[Music Alchemist]

"Neutral" is kind of a loaded word, especially due to the debate over which compensation curve is the most neutral one.
 
I think we can all agree that "more accurate" is a more accurate (*snicker*) term to use.

 

[castleofargh]

neutral exists, it's whatever we get flat on some graph

.
 
with a DAC electrically flat is neutral. with speakers it's the same, even if we don't get perfect flat, having flat speakers will sound "neutral" to us, as in real life usual sound.
so I believe there is the exact same neutral for our ears with a headphone. it's probably an individual quest at some point, but if my headphones could sound to me like they have the same signature as good speakers, I'm very confident that would be my neutral.
accurate may involve more than just frequency response no?

 

[Music Alchemist]

  neutral exists, it's whatever we get flat on some graph

.
 
with a DAC electrically flat is neutral. with speakers it's the same, even if we don't get perfect flat, having flat speakers will sound "neutral" to us, as in real life usual sound.
so I believe there is the exact same neutral for our ears with a headphone. it's probably an individual quest at some point, but if my headphones could sound to me like they have the same signature as good speakers, I'm very confident that would be my neutral.
accurate may involve more than just frequency response no?

 

 
Where did you get the idea that a flat response with loudspeakers or headphones looks like a flat line?
 
Because any of the curves (diffuse-field, free-field, Harman, etc.) that represent what our ears hear are not a flat line.
 
And neutral-ish headphones (or any headphones for that matter) most certainly do not measure as a flat line.
 
You have to compensate for human hearing. Transducers (headphones and speakers in this case) are not like electronics in this respect. I'm surprised someone of your stature was not aware of this. The only time it would look like a flat line is if you were comparing post-equalization measurements to a compensation curve and managed to get them to align perfectly.
 
Some reference links:
http://www.innerfidelity.com/headphone-data-sheet-downloads
http://www.innerfidelity.com/headphone-measurements-explained
http://www.innerfidelity.com/content/harman-researchers-make-important-headway-understanding-headphone-response
http://www.innerfidelity.com/content/headphone-target-response-curve-research-update
 
But I may have misinterpreted you. I thought you meant a literal flat line on a graph. If not, disregard my rant. ^_^
 
In any case, I'm curious which compensation curve you feel is the neutral one for headphones, and why.

 

[RRod]

   

 
Where did you get the idea that a flat response with loudspeakers or headphones looks like a flat line?
 
Because any of the curves (diffuse-field, free-field, Harman, etc.) that represent what our ears hear are not a flat line.
 
And neutral-ish headphones (or any headphones for that matter) most certainly do not measure as a flat line.
 
You have to compensate for human hearing. Transducers (headphones and speakers in this case) are not like electronics in this respect. I'm surprised someone of your stature was not aware of this. The only time it would look like a flat line is if you were comparing post-equalization measurements to a compensation curve and managed to get them to align perfectly.
 

 
It's all about where you're talking about the measurement. When we say "flat" speakers we mean "measured at a given distance and orientation in an anechoic environment." Once you put a reflective room into the mix then you compensations to worry about, but the paradigm of "flat from dac to amp to speakers" is a good one, because it removes variables. Similarly we can define a flat headphone as "matches the at-ear-canal response of flat speakers in a subjectively 'nice' room." A bit more fuzzy, but still something measurable. I don't think argh was trying to say we want cans to measure ruler flat in and of themselves (if I can put English words into his mouth).

 

[Music Alchemist]

  It's all about where you're talking about the measurement. When we say "flat" speakers we mean "measured at a given distance and orientation in an anechoic environment." Once you put a reflective room into the mix then you compensations to worry about, but the paradigm of "flat from dac to amp to speakers" is a good one, because it removes variables. Similarly we can define a flat headphone as "matches the at-ear-canal response of flat speakers in a subjectively 'nice' room." A bit more fuzzy, but still something measurable. I don't think argh was trying to say we want cans to measure ruler flat in and of themselves (if I can put English words into his mouth).

 
Okay, as long as he didn't mean a literal flat line.

 

[Lunatique]

Okay, I just edited the original post with as many Youtube links to the songs as I can find. I tried to find the highest resolution versions on Youtube too, whenever I could.

 

[castleofargh]

 

  It's all about where you're talking about the measurement. When we say "flat" speakers we mean "measured at a given distance and orientation in an anechoic environment." Once you put a reflective room into the mix then you compensations to worry about, but the paradigm of "flat from dac to amp to speakers" is a good one, because it removes variables. Similarly we can define a flat headphone as "matches the at-ear-canal response of flat speakers in a subjectively 'nice' room." A bit more fuzzy, but still something measurable. I don't think argh was trying to say we want cans to measure ruler flat in and of themselves (if I can put English words into his mouth).

 
Okay, as long as he didn't mean a literal flat line.

yeah what RRod said.  electrically flat headphones would sound like crap of course ^_^. I meant to say that neutral could be seen as the notion of having a flat line somewhere on some graph. you just have to stop and say "this is the new flat". we can just take whatever signature, calibrate with that and now that signature is "neutral"

in that given compensation, and would give a flat line when measured that way. 
it would have little to do with my perceived balance of frequencies which would be my "neutral", but a totally different thing. that was a little ironical. sorry for the lost in translation.

 

[spruce music]

  yeah youtube is very fine. I'm sorry that it makes more work for you but the law is the law.
 
thank you for the EQ, I'll try that tomorrow seriously(it's 1.30AM for me right now ^_^)
 
the loudness has an impact for sure. I imagine it would work better at louder volumes as the FM loudness curve flattens out a little, but loud test tones, that's not a lot of fun.

 
still I do not feel like equal loudness is the same as perceived neutral. and you're making people do equal loudness(and then correct it with actual music). that's why I said it's good for spikes, but not to get overall neutral sound IMO.  I doubt anything but a neutral reference from good speakers can let people achieve neutral on headphones. but again I'm good with your method, it will improve some things on most headphones, and let people not too far off of neutral, so they can fine tune by a few dbs later on with music. or maybe apply a default fletcher munson compensation? we all have our own + the loudness stuff, but the general direction would still be good I guess.
 
when I EQ my music, the ultimate judge is.... November rain

on this video starting at around 7mn, when my EQ is right, 1/ the guitars on both sides feel balanced, and 2/ the third one in center must dominate clearly the other 2, and 3/not be overwhelmed by the percussion. then if the voices are good it's even better, but if I get it right with the guitars+percussion, I know music will sound nice. ^_^
I wish my reference song was the Al Jarreau version of blue rondo a la turk, or some fancy Pink Floyd stuff, or the usual Bach organ stuff. but nope, the one song that does it for me is November rain

 

All of that is why I think the log sweep would benefit being pre-EQ'd for Fletcher Munson.  Pick the 70 or 80 phon version.  It will not be perfect, but will be very much in the right direction.  Otherwise, as you have been saying castleofargh inexperienced people will end up with their personal loudness contour.  Taking most of this out by EQ of the sweep will put them much closer from the beginning to flat response which most definitely is not the loudness contour curve. 

 

[Music Alchemist]

  yeah what RRod said.  electrically flat headphones would sound like crap of course ^_^. I meant to say that neutral could be seen as the notion of having a flat line somewhere on some graph. you just have to stop and say "this is the new flat". we can just take whatever signature, calibrate with that and now that signature is "neutral"  in that given compensation, and would give a flat line when measured that way. 
it would have little to do with my perceived balance of frequencies which would be my "neutral", but a totally different thing. that was a little ironical. sorry for the lost in translation.

 
I see. So we're on the same page after all! The only problem is which compensation curve to use. That is a heated debate in itself.

 

[Lunatique]

 
yeah what RRod said.  electrically flat headphones would sound like crap of course ^_^. I meant to say that neutral could be seen as the notion of having a flat line somewhere on some graph. you just have to stop and say "this is the new flat". we can just take whatever signature, calibrate with that and now that signature is "neutral"

in that given compensation, and would give a flat line when measured that way. 
it would have little to do with my perceived balance of frequencies which would be my "neutral", but a totally different thing. that was a little ironical. sorry for the lost in translation.

I'll clarify what I meant exactly by neutral/accurate.
 
In my studio, at my listening position, the measuring mic of the ARC System 2 takes measurements of my speaker system (full-range mastering grade 2.1 system that includes the Klein+Hummel O 300Ds and Neumann KH805) and the effects of room mode due to any acoustic issues that's not fully taken care of by the acoustic treatments in my studio. Also, the exact placement of my speaker system was the result of an on-going consultation with Neumann's product portfolio manager of studio monitor systems (as some of you might know, Sennheiser is the parent company of Neumann). I'd take measurements and then show the screenshot to him, as well as take photos of the studio. He'd give me a set of tips and instructions, and I'd carry them out (tweaking the hardware settings on the speakers themselves, testing various positions with measurements, do the famous "subwoofer crawl," record a $hitload of data and calculating them, etc., then take measurements and photos again, and repeat the process over and over. The entire process took over two months of refining, and finally I was able to achieve the most neutral/accurate response I have ever gotten without any corrections. Then to finish it off, I take one final measurement at the listening position with the ARC System 2, and it corrects the frequency response and time-domain issues and gives me perfect 20Hz-20KHz frequency response and precise imaging with no time-domain issues.
 
So that is my golden standard for neutrality/accuracy--a full-range mastering grade 2.1 sound system that does 18Hz-20KHz and is measured perfectly flat from 20Hz-20KHz at the listening position, in an acoustically treated studio. 
 
With headphones, my goal is always to try to get as close to that as possible. So, no, I would not want a perfectly flat line for a headphone's frequency response, and I mentioned this in the original post when I posted those links to InnerFidelity's articles on understanding headphone measurements and the Harman Target Response Curve.

 

[Music Alchemist]

With headphones, my goal is always to try to get as close to that as possible. So, no, I would not want a perfectly flat line for a headphone's frequency response, and I mentioned this in the original post when I posted those links to InnerFidelity's articles on understanding headphone measurements and the Harman Target Response Curve.

 
Yes, but he did confirm that he wouldn't want a flat line for the actual measurements either; he was just referring to aligning the equalized response of the headphone with a compensation curve, then using that compensation curve as the new flat line, so to speak, similar to how Golden Ears and other projects do it.
 
So since you got such an accurate response for the speakers...would you say that headphones sound like that when equalized to the Harman curve?
 
Various other companies and individuals use their own compensation curves. I'm very interested in learning which ones are the most accurate.

 

[Lunatique]

   
Yes, but he did confirm that he wouldn't want a flat line for the actual measurements either; he was just referring to aligning the equalized response of the headphone with a compensation curve, then using that compensation curve as the new flat line, so to speak, similar to how Golden Ears and other projects do it.
 
So since you got such an accurate response for the speakers...would you say that headphones sound like that when equalized to the Harman curve?
 
Various other companies and individuals use their own compensation curves. I'm very interested in learning which ones are the most accurate.

I was responding to them wondering if I meant I wanted totally flat response for headphones. 
 
Speakers will always sound better than headphones. The reason is because soundwaves from speakers must travel in the air and interact with the molecules in the air and excite them before entering our ears, as well as interact with the acoustic space of your listening space, and also the actual vibrations that our body can sense. That simple fact makes speakers far more visceral, dynamic, alive, dimensional, and have wider and deeper soundstage. It's possible to get headphones to sound like speakers in terms of frequency response, but in other aspects it's extremely hard or maybe even impossible to get them to match speakers. I personally don't think it's possible without some really high-quality and complex DSP magic, such as using real recorded impulse response from actual speakers in a listening space, and even then it probably won't be 100%.
 
I would pick my speaker system over any headphone, any day, no contest, especially after I've spent the past 8 years tweaking and perfecting that system (including the actual studio space) to the absolute pinnacle of performance. I don't know if you guys realize this, but it's extremely rare for those in the head-fi community to achieve perfect neutral/accurate sound with a speaker system that goes from 18Hz to 20KHz. Those who can afford it often don't know about or use room correction, and those who know about room correction often can't afford it. Those who do know about it and can afford it often don't know enough about acoustics and room modes to know where exactly to place the speakers/subwoofer in their listening space to avoid significant nulls and peaks caused by room modes. Even if they did research about it online, there is a lot information missing that you really have to consult the experts extensively in order to achieve this goal. For me, it took prolonged correspondences with an engineer at Klein+Hummel and a product manager at Neumann to finally get my speaker system to sound this amazing. 
 
Essentially, I would never put on a pair of headphones unless I absolutely need to. And when I do, I want the headphone to sound as close to my speaker system as possible. Even if I know I'll never get it to sound close enough, it's still going to sound much better than if I don't try to correct the headphone at all. 

 

[spruce music]

The problem with EQ of headphones is what is the reference?  With speakers you can measure it with calibrated gear.  With headphones they interact with each individuals ear and fit so much that won't work (at least for DIY efforts).
 
Now just as a thought experiment suppose we had what we already knew was a truly flat phone for the person wearing it. If we play a log sweep it will not sound of even loudness.  If the user then adjusts for even loudness every adjustment is moving away from flat accurate response.  The user ends up with his own personal equal loudness curve and not accuracy. 
 
Now let us flip it around suppose we also know not only our phone is truly accurate, but the equal loudness contour for the user.  If we EQ the log sweep for the users equal loudness curve and play it back over the flat phone the user will hear even loudness across the band.  Now if the user swaps out his headphones for another model not flat in response, he could play the EQ'd for equal loudness log sweep and adjust the second headphone until equal perceived loudness across the band was achieved.  This second phone would now be EQ'd in a way to provide flat response. 
 
Now I would suggest Fletcher-Munson curves are known to be generally close on average for the majority of people.  They aren't perfect for every person especially at high frequencies with the effects of aging. But using the curve for something like the 80 phon level should be at least in the general ballpark.  So doing EQ to your log sweep based upon that is much more than likely to be closer to flat that not doing it. Then once you have EQ for your phones as close to flat, you can use the suggested Harman EQ or others.  I have in the past used 3 db/decade slopes starting at 200 hz.  So down -3db for 2khz and -6 db for 20 khz.  Actually you should use the latest ISO standard for this.
 
http://www.sengpielaudio.com/Acoustics226-2003.pdf
 
Another possibility is using your speakers as a reference if they are better than your phones in flatness.  Or using your speakers to develop your own equal loudness curve which then can be applied to pre EQ the log sweep to adjust headphones for equal perceived loudness.  It isn't perfect, but it is a reference to adjust off of which is far better than nebulous adjusting to taste with no reference touchstone. 
 
It is possible using the Harman Headphone target response curve is actually better to EQ the log sweep.  I have not tried that.  If the right curve it should work more or less. EQ the log sweep with the Harmon curve and adjust for equal loudness.  I suspect this might end up too bright if adjusted to equal loudness.  Apply some tasteful tilt and it should be good.

 

[castleofargh]

yup, the ideal situation is to have a nice flat speaker system calibrated in a nice room. then we have one reference and there are ways to try and get something similar (like following @Joe Bloggs tuto that does involve equal loudness contour for 2 devices as a way to know the FR response difference between the 2).
 
but without a reference headphone/speaker, it's pretty hard to know where we're going just with loudness. it's a lot of trial and error before getting something that does sound clear everywhere. now there seem to be some common grounds between hearing, measurements, taste, etc. at least for maybe 200hz to 7khz ou 8khz when I get something I like, I can just take another headphone, EQ it from measurements and get something not too far off in that range.
where I get totally lost is when I try to EQ an open back and a closed back, or for IEMS, a vented dynamic driver and a sealed balanced armature driver. measurements seem to be very different to what I'm hearing from each techs(and that's the only reason why I keep EQing by ear for some IEMs).
like right now I'm in love with the fiio EX1(AKA titan1), I find that it's a very good EQ material, the low end doesn't drop much if at all, and the trebles are too strong, so easy to just turn down without having to fear distortions. but it's not isolating so it's use is limited sadly.  I measured it with the vibro veritas coupler and got a flat mid and low end response, something rather close to what I got from measuring my hf5.  but here is the thing, I always feel like I need more low end on the hf5, when I find the EX1's low end to be amazing...

I'm guessing the coupler doesn't react the same way if it's sealed of not, or other acoustic reasons I don't understand. but I clearly reach some situations like that where measurements just don't make for a proper comparison at all.
 
and I don't have a good pair of speakers I can calibrate and rely upon as reference. until I move to some other place, speakers are a no go for me.

 

[Music Alchemist]

  I was responding to them wondering if I meant I wanted totally flat response for headphones. 
 
Speakers will always sound better than headphones. The reason is because soundwaves from speakers must travel in the air and interact with the molecules in the air and excite them before entering our ears, as well as interact with the acoustic space of your listening space, and also the actual vibrations that our body can sense. That simple fact makes speakers far more visceral, dynamic, alive, dimensional, and have wider and deeper soundstage. It's possible to get headphones to sound like speakers in terms of frequency response, but in other aspects it's extremely hard or maybe even impossible to get them to match speakers. I personally don't think it's possible without some really high-quality and complex DSP magic, such as using real recorded impulse response from actual speakers in a listening space, and even then it probably won't be 100%.
 
I would pick my speaker system over any headphone, any day, no contest, especially after I've spent the past 8 years tweaking and perfecting that system (including the actual studio space) to the absolute pinnacle of performance. I don't know if you guys realize this, but it's extremely rare for those in the head-fi community to achieve perfect neutral/accurate sound with a speaker system that goes from 18Hz to 20KHz. Those who can afford it often don't know about or use room correction, and those who know about room correction often can't afford it. Those who do know about it and can afford it often don't know enough about acoustics and room modes to know where exactly to place the speakers/subwoofer in their listening space to avoid significant nulls and peaks caused by room modes. Even if they did research about it online, there is a lot information missing that you really have to consult the experts extensively in order to achieve this goal. For me, it took prolonged correspondences with an engineer at Klein+Hummel and a product manager at Neumann to finally get my speaker system to sound this amazing. 
 
Essentially, I would never put on a pair of headphones unless I absolutely need to. And when I do, I want the headphone to sound as close to my speaker system as possible. Even if I know I'll never get it to sound close enough, it's still going to sound much better than if I don't try to correct the headphone at all. 

 
Loudspeakers may sound better than headphones for some people, but not me. Sure, loudspeakers can do plenty of things headphones can't, and they do sound big and impressive...but also distant and impersonal. I don't connect with the music at all with them. I prefer headphones nearly 100% of the time, partially due to the more intimate presentation that makes me feel one with the music.
 
Here's one thing that can make headphones sound more like speakers, but it's a little pricey: http://www.smyth-research.com
 
If you don't mind divulging, how much did it cost to do all that?
 
Also, I'm still curious about my question relating to whether Harman-equalized headphones sound like your speaker system in terms of frequency response.