Do Objective Headphone Measurements Correlate to the Audiophile's Subjective Experience? - Page 13

Flat for a speaker, not quite flat for a headphone. Headphones need a reduction in upper mids and treble to sound more natural. The ironic thing is that even though the upper mids on the graphs for the denons show them to be right in line, in actual listening they're not very balanced with the rest of their sound.

 

Sorry, I meant flat as in a flat line! =)  I've seen at least four different theories on what "flat" in a headphone is, different kinds of lines, even a mirror image of an average equal loudness contour called the golden line of FR used in the UE monitors and the UM Miracle.

 

Of course, there is no real consensus as to what the compensation curve SHOULD be, so take this with a few grains of salt. The top trace in the D-5000 graph is "compensated" (that is, adjusted by Innerfidelity's compensation curve), so flat should be proper. At least if you believe that flat IS proper...

 

Many people believe that recordings are, by nature, recorded too closely and have too much treble so that playback should be rolled off in the treble in any case. I would not argue against this.

No if anything the uncompensated curve is more in line with what our ears actually hear, the compensated graph is compensated to what a machine hears. Our ears have ridges that amplify direct sound in the upper mids and lower treble-- speakers have the advantage of being faced far away from our ears. Tyll mentions this in his videos about measuring headphones and he comments about what he believes to be ideal fr graphs in his 'comparing world class headphones' article when analyzing the measurements. Of course everyone's ears are shaped differently so there will be no definitive neutral for headphones, just as certain hrtf filters like Dolby Headphone and ultrasone's surround sound doesn't work best with everybody. Keep in mind im only talking hrtf and not even getting into loudness eq curve.

You do have a point about recording distance but these are two different things.
Edited by TMRaven - 7/1/12 at 8:39pm

AFAIK the compensation for IF is the HATS' own HRTF. The raw is like it says, raw. And the arguments that can be started by talking about what "flat" actually is are absolutely virulent.

I was alluding to that Sean Olive presentation many pages back, and it's interesting to see purrin's results mirror that. However that doesn't mean a whole lot by itself - so we can say that the majority of listeners will prefer X, but does that tell us anything about what Tom up the block really wants? Sure, it's nice to know what is most appealing or what have you, but it doesn't let you predict as much as some objectivists would insist.

Finally, the comment that Olive made about "headphones today are where speakers were 15 years ago - before we started measuring" seems a bit out of place (and don't think I'm trying to rip on Olive here) - I think speakers overall have trended towards "flat" or "listenable" for decades, simply because it sells (go dig your grandparents' speakers out of the attic if you don't believe me - even those big towers from the 1970s and 1980s are relatively flat sounding and relatively listenable, assuming they're in proper condition). Headphones, by contrast, seem to have a few different things going on; there's the bassy/rapper can genre that's relatively new, the "mid-bass ooooh" cans (I remember someone talking about Bose headphones having this feature in their FR, and I think a lot of "hi-fi" cans have this feature to some extent as well), and then the super-duper bright-as-the-sun "audiophile" cans. Yeah I'm generalizing and exaggerating, but think about it - thats three VERY different paradigms. Speakers on the whole don't seem to target three completely unrelated moving references. I don't think measurements are going to change that either; beyond letting a bunch of people go on crusade and try to tell everyone else what they should like (which a lot of people have done based off of Olive's work when it comes to speakers, which is a real shame to see).

I think measurements are neat, but I like Tyll's summary of it - measure what you like, get an idea of what you like looks like on a graph, and then compare that to measurements of other things and see if it jives. If it's completely different, you might not like that other thing, if it's fairly similar, you will probably like that thing. That's a bit different than trying to establish concrete rules to tell everyone what they should want to listen to.

Just my .02.

Thats not to say that multimedia packaged speakers either don't come with their own midbass hump (aka audio engine) or come with pathetic excuses for subs that are just midbass modules.

There's definitely far less deviation in speaker sound as a whole, though. I'd like to see Grado make a pair of speakers.

+1. And Grado has made a few pairs of speakers over the years.

Yah, and they're really good and neutral... kind of like a HP 1000, according to Larryminator and a few other sources.

 

 

Pictures of the two models existing (regular and "JR") in my thread :)

Which most accurate Frequency response graph between Headroom, purin changstar, and innerfidelity ?

 

Accuracy is really hard to come by in this arena. Repeatability is more important, and in that regard HeadRoom's graphs (mostly the early measurements) may not be as good.

 

InnerFidelity's and Purrin's are probably both good, Purrins have CSDs, which is really cool.

 

I have the largest database of measurements (I think I have more than HeadRoom now), which is helpful because there are more to compare.

 

The Golden Ears are probably good too.  But the ones at the more mainstream outlets (Um, Sound and Vision? Headphonereviews? Can't remember atm) are a bit sketchy it seems to me.

In my subjective opinion my current favorite sites for measurements goes in this order:

 

1. Goldenears (larger collection than changstar)

2. Changstar, seems to be pretty consistent measurements but the selection not as good as goldenears

3. Innerfidelity, good selection of interesting cans but the fluctuations in the upper mids/lower-highs, around 4-8kHz range, makes me put it 3rd, my mind automaticly interpolates the innerfidelity graphs a bit when seeing those typical 5-7kHz dips. The question is exactly how much to "interpolate" though and how does this vary depending on headphone ^^

Edited by RPGWiZaRD - 3/6/13 at 10:23am

I think if this question was posed another way results would be more definitive.

 

1.)  What headphone has more bass?

2.)  What headphone has more treble?

3.)  What headphone sounds the most neutral?

 

With those direct questions I highly suspect that the results would directly correlate to the graphs.  The issue with subjective measurements is that the question / test is rarely specific enough to extract a good result.  It is much like asking what headphone do we all think sounds the "best".  What the heck does best mean to many people?  I fully expect as many different results as there are people, but if you ask one of the questions above regarding two headphones the subjective results will be much more reliable.

 

I have always used the graphs to tell me what I should be hearing.  If I am not hearing what the graphs show then either the graphs do not correlate to my gear or there may be something wrong with my setup ( impedance, bad cable, poor DAC, low output current / voltage ).

 

I think most of us will agree that having both the graphs and direct subjective results is extremely valuable.  I respect both sides.