[Alondite]

 I'm sure we all know about the response graphs over at Headphone.com, but we also know that what we see in those graphs may not be representative of what we are hearing.
 
My question is: what causes this difference? Is it merely a perceptual difference? That is, is the difference between how it's supposed to sound and how it actually sounds just our perception of the sound? Is it ear-shape? Or are these graphs just plain wrong? Can perceptual hearing differences and structural ear differences also be why people have such night-and-day different opinions and analysis of how a headphone sounds?
 
These graphs give us an objective measurement of the headphone's true performance (provided that the tests minimize testing error, which I'm sure they do), but how do they directly translate into sound? Or rather, how are they -supposed- to translate into sound? I don't have much experience with different headphones, so I'm not entirely sure how they are "supposed" to sound. I'm not totally clueless, but not as informed as I would like to be.
 
This all stems from the fact that I'm currently entering the world of audiophilia and higher-end sound equipment. On top of being a videophile it gets pretty expensive, and it's not like I can just go somewhere and test every headphone I might ever consider buying. As such, I would like to know if I can use the response graphs relative to what I have heard myself and from first-hand accounts to make my purchasing decisions.
 
More than anything, I suppose, I want to know how these graphs directly translate into sound. I know what a frequency response graph is showing me, I know what a square wave is, what I don't know though is how what is shown in these graphs is supposed to translate into sound. Whether it actually does or not (which as I alluded to before could just be perceptual differences), what do these graphs say about the sound? I am, however, a bit clueless on the distortion graph, though the results seem to translate more subjectively.
 
I'll use some examples.
 
 
.
 
Here is the frequency response graph for the Klipsche X-10i. This graph tells me that this particular pair of phones has slightly elevated (over the +3-4 dB Headroom states as being the target), but relatively even bass response, and then rolls off in the highs. Headroom says that about 8-10 dB of roll off is appropriate, but these roll off about 30 dB even before 20 kHz. That seems like a lot to me, and I would think that the high end would suffer a bit. Is that true? Also, what about the peaks and valleys in the 1 kHz+ range? I've heard those are desirable, but why? I've also heard that they are only desirable to a certain extent, so how much is too much? Examples would be preferred.
 
Here is another example.
 
:
 
These seem very balanced to me, but I'm wondering why the bass starts low at 10 Hz and climbs through about 40 Hz. Is this intentional? And why would it be? I would think that these would have pretty weak bass.
 
 
Here is one of the square wave graphs available. 500 Hz tells me that it's more representative of mid-high response, and I know what this graph would look like ideally. What I don't know though, us how a particular result translates into sound.
 

 
The Senn HD800s are by virtually all accounts some of, if not the best headphones ever made. However, I don't believe that this is the best 500 Hz SW graph I've ever seen. It does resemble a square wave, but the horizontal surfaces aren't particularly flat. What does this mean about the sound? Why is there a spike on the front edge of each horizontal portion, and how would it sound? If this is a good graph, than what would a bad one look like?
 
The 50Hz graphs seem even more varied and difficult to read. Some of them look virtually identical to a square wave, and some of them look nothing at all like a square wave Yet, some of the headphones that yield stranger-looking graphs still sound excellent.
 
 

 
Take this one for example. According to Headroom, these are an excellent pair of headphones for their price, but what is with this graph? It looks more like a spiky sine wave than a square wave. How would something like this translate into sound? What's with the giant spike in front of each segment, and then the following "sine wave." What does that mean about the sound? I've seen other headphones in this price range that have a much more accurate looking square wave graph.
 

 
Though the X-10i are not in the same price range, this looks like a fantastic square wave graph. Am I correct in my assumptions? And how does a "correct" square wave graph translate into sound?
 
 

 
Here is another graph. This aren't quite ideal, but they still look pretty decent. I'm wondering though, about the curved front edge and the later peak. What does this mean about the sound? I also notice that many of the headphones taper off a great deal after their peak, what does that say about their sound?
 
I'm not going to bother posting a distortion graph, because quite frankly I'm not sure what I would ask about it. However, any information would be greatly appreciated. Impedance and Isolation graphs are simple enough though, so I don't have any questions about those.
 
I know this is asking a lot all at once, but I'm really just trying to learn. Thanks in advance.
 

 

[BlackbeardBen]

Welcome - and wow - what a first post!
 
You ask many excellent questions, and they're stated exceptionally well.  In fact, so well they sound like the sort of questions I would expect as the foundation for serious scientific research into the topic!
 
I'll leave others more qualified to attempt to answer them though.

 

[maverickronin]

I'm not an expert on the ins and outs of the various measurement techniques used by Headroom but I can give you a few basic answers.
 
The first thing to keep in mind is that headphones are much harder to measure properly than speakers are because there is no agreed upon standard for what constitutes a perfect headphone.  A speakers sits in a room like an instrument and simply has to play what is fed as accurately as possible.  Soundstaging is handled in the studio by mixing for speakers as a target.  Headphones on the other hand usually play music which was mixed for speakers.  On top of that, the room the speakers sit in as well as your face, ears, and torso EQ the sound that actually reaches your ears.  That physically 'emulates'  the sound of actual instrument or performers sitting in real space.  Headphones have to design all that into the transducer and its enclosure.  Even discounting or averaging individual characteristics, no one can agree up a standard about how it should sound.  The Headroom graphs are "corrected" for diffuse field EQ, which is the closest thing there is to a standard.  Subtract 2 from the graphID in the URL to see the raw data.  Here's how the D7000s look uncorrected.
 

 
You said they looked bass-light before, but now they're nothing but treble.  Your new here so probably don't know, but most people on head-fi would consider these bass monsters.  Doesn't make much sense does it?  I wouldn't call them bass monsters myself, but I'm not much of a fan of DF EQ.  Most people around here are treble-heads though.
 
At this point you're probably wondering if there graphs are good for anything at all.  To me, they are very useful, but not on their own.  I've owned and sold a fair number of models which have been measured on Headroom which is where their real value is.  I can correlate something I've heard with Headroom's data on that model and use that to evaluate anything new that they measure with the same methodology.  That's the real value of these graphs.

 

[Uncle Erik]

If you want to complicate things some more, the shape of your ears and ear canals also affect how you hear a headphone.

Then there are the impedance curves. Most headphones change impedance over the frequency range. More complicated still is that there's a wide variety of headphone amps - really diverse. All sorts of solid state and hybrid amps, not to mention tubes. The tubes can vary a lot because you don't need big power tubes, and there's a lot of different preamp/low power tubes. Some of the tubes aren't that linear, either, but sound pleasing.

You're beginning to see just how many combinations there are. Don't despair. Also, you're not going to keep your first pair forever so don't kill yourself over the choice or go all in with one of the expensive ones.

I usually tell people to start out with one of the classic headphones. Something like a Grado SR-60 or a Sennheiser HD-600. Those are great baseline headphones thousands of members have used. Depending on how you feel about them, you'll know which step to take next. If you think a Grado is too colored or piercing, then you'll know where to look next. If you love the HD-600 but want more clarity, you'll have options.

Also, make an effort to attend a meet. Everyone is welcome and you don't need to bring gear. But bring your music and try out as many things as possible. You'll get a much better idea of what's out there. You aren't going to find your ideal headphones right away. Though you will have a lot of fun discovering the different sounds and getting to know the cast of characters around here.

 

[Alondite]

Quote:

Welcome - and wow - what a first post!
 
You ask many excellent questions, and they're stated exceptionally well.  In fact, so well they sound like the sort of questions I would expect as the foundation for serious scientific research into the topic!
 
I'll leave others more qualified to attempt to answer them though.

Thank you! I believe in making a good first-impression, so I tried to put some real thought into my first post. If I had the time and the resources I would certainly consider doing some in-depth research on the topic.

 
Quote:

I'm not an expert on the ins and outs of the various measurement techniques used by Headroom but I can give you a few basic answers.
 
The first thing to keep in mind is that headphones are much harder to measure properly than speakers are because there is no agreed upon standard for what constitutes a perfect headphone.  A speakers sits in a room like an instrument and simply has to play what is fed as accurately as possible.  Soundstaging is handled in the studio by mixing for speakers as a target.  Headphones on the other hand usually play music which was mixed for speakers.  On top of that, the room the speakers sit in as well as your face, ears, and torso EQ the sound that actually reaches your ears.  That physically 'emulates'  the sound of actual instrument or performers sitting in real space.  Headphones have to design all that into the transducer and its enclosure.  Even discounting or averaging individual characteristics, no one can agree up a standard about how it should sound.  The Headroom graphs are "corrected" for diffuse field EQ, which is the closest thing there is to a standard.  Subtract 2 from the graphID in the URL to see the raw data.  Here's how the D7000s look uncorrected.
 

 
You said they looked bass-light before, but now they're nothing but treble.  Your new here so probably don't know, but most people on head-fi would consider these bass monsters.  Doesn't make much sense does it?  I wouldn't call them bass monsters myself, but I'm not much of a fan of DF EQ.  Most people around here are treble-heads though.
 
At this point you're probably wondering if there graphs are good for anything at all.  To me, they are very useful, but not on their own.  I've owned and sold a fair number of models which have been measured on Headroom which is where their real value is.  I can correlate something I've heard with Headroom's data on that model and use that to evaluate anything new that they measure with the same methodology.  That's the real value of these graphs.

I didn't know about finding the raw data graphs, thank you! Could their "monster" bass have something to do with their 50 Hz square wave graph?
 

 
 
Again, I'm not quite sure how to read this, or how it translates into sound, but it seems to me that there is a lot of pressure that accompanies the frequency shift. That is what those large spikes mean, right? Logically it makes sense to me, but I'm not entirely sure. I wish there was somewhere I could generate a square wave, and then tweak it to make it resemble one of these graphs so I could see how it sounds.
 
 
Quote:

If you want to complicate things some more, the shape of your ears and ear canals also affect how you hear a headphone.

Then there are the impedance curves. Most headphones change impedance over the frequency range. More complicated still is that there's a wide variety of headphone amps - really diverse. All sorts of solid state and hybrid amps, not to mention tubes. The tubes can vary a lot because you don't need big power tubes, and there's a lot of different preamp/low power tubes. Some of the tubes aren't that linear, either, but sound pleasing.

You're beginning to see just how many combinations there are. Don't despair. Also, you're not going to keep your first pair forever so don't kill yourself over the choice or go all in with one of the expensive ones.

I usually tell people to start out with one of the classic headphones. Something like a Grado SR-60 or a Sennheiser HD-600. Those are great baseline headphones thousands of members have used. Depending on how you feel about them, you'll know which step to take next. If you think a Grado is too colored or piercing, then you'll know where to look next. If you love the HD-600 but want more clarity, you'll have options.

Also, make an effort to attend a meet. Everyone is welcome and you don't need to bring gear. But bring your music and try out as many things as possible. You'll get a much better idea of what's out there. You aren't going to find your ideal headphones right away. Though you will have a lot of fun discovering the different sounds and getting to know the cast of characters around here.

 
Yeah I've noticed how much accessory equipment there is, and the different results you can get from mixing and matching equipment. It is quite daunting at the moment, but I'm sure I'll come to grips with it before too long.  As for a meet, that sounds like something I would definitely like to do. I'm near Syracuse NY, so I'm relatively certain I would be able to make it to one sometime.
 
At the moment I have a pair of Klipsch Image S4 that I've had for about 2 years now. At the time they seemed like one of the better options in my price range, so that's what I went for. The more I listen to them though, the more I notice the flaws in the sound. I'm getting to a point now where I feel it's about time to upgrade.
 

 

[maverickronin]

Quote:

I didn't know about finding the raw data graphs, thank you! Could their "monster" bass have something to do with their 50 Hz square wave graph?
 
 
 
 
Again, I'm not quite sure how to read this, or how it translates into sound, but it seems to me that there is a lot of pressure that accompanies the frequency shift. That is what those large spikes mean, right? Logically it makes sense to me, but I'm not entirely sure. I wish there was somewhere I could generate a square wave, and then tweak it to make it resemble one of these graphs so I could see how it sounds.

AFIK, that spike is basically the driver overshooting its target, which is quite common.  It doen't have much to do with percieved bass as far as I can tell since even 'phones which are very light on bass have similar overshoots.
 

 
And there are also 'phones with more neutral bass like these two which have very different levels of overshoot.
 

 
So I don't really know why people would say that about them.  I've owned the D2000s and D5000s and heard the D7000s at meets.  I didn't think they had excessive bass though it wasn't very "tight" down deep.  I think some people are just alergic to bass and others seem to intrpret any bass that's not not near perfect as excessive and since almost nothing is near perfect they don't like any bass at all.

 

[rroseperry]

I think this is also a useful thread, especially since Tyll is going to write up explanations of what the graphs suggest about the different headphones.

 

[Alondite]

Quote:

Quote:

AFIK, that spike is basically the driver overshooting its target, which is quite common.  It doen't have much to do with percieved bass as far as I can tell since even 'phones which are very light on bass have similar overshoots.
 

 

I wonder if perhaps the speed and extent of drop-off in the SA5000s have anything to do with their light bass?  I'm assuming they are light in bass, which is why you used them in comparison.  The D7000 seems to drop off more gradually, so maybe the more sustained pressure means something? There has to be some way to translate these graphs into sound.
 

 

[Tyll Hertsens]

Quote:

 There has to be some way to translate these graphs into sound.
 

 
Workin' on it.
 
 

 

[Alondite]

So I've heard! Have you found anything yet? And if so is there anywhere I could see these findings?
 
On an unrelated note, I loved your review of the Beats Solos. Those things are wildly popular and it bothers me because they really don't sound good.

 

[Tyll Hertsens]

  So I've heard! Have you found anything yet? And if so is there anywhere I could see these findings?
 
On an unrelated note, I loved your review of the Beats Solos. Those things are wildly popular and it bothers me because they really don't sound good.

 
 
 
 
They will start showing up in the first two links on this page in about a week or so.
 
 
 
And, yeah, bothers me too.