[Bredin]

So I heard you like smoothing!

 

[b0dhi]

There seems to be some confusion here. The graphs you see on HeadRoom's site show the raw response of these headphones. Sennheiser's graphs (contrary to the axis labels) show the equal-loudness corrected response. Both are "correct", it's just that they are showing different things. In Senn's graph, the raw response is compared to some equal-loudness curve and the difference between them is the line you see. Unfortunately Senn isn't explicit about where they get their equal-loudness curve.

 

[Bullseye]

^Fletcher and Munson curves might be the ones they are using. There are loads of equal loudness curves to take as reference.

However it still strikes me that the graph only shows till 12k instead of 16k or 20k...

Then it is curious to see they are measuring using a DF microphone. Would like to see how the AKG K240DF measure against these.

EDIT: For the ones interested in knowing more about equal lodness curves you can read these study: http://www.zainea.com/equalloudnesslevel.pdf

 

[Bredin]

Quote:

Originally Posted by b0dhi /img/forum/go_quote.gif There seems to be some confusion here. The graphs you see on HeadRoom's site show the raw response of these headphones. Sennheiser's graphs (contrary to the axis labels) show the equal-loudness corrected response. Both are "correct", it's just that they are showing different things. In Senn's graph, the raw response is compared to some equal-loudness curve and the difference between them is the line you see. Unfortunately Senn isn't explicit about where they get their equal-loudness curve.

But sennheisers graph is still VERY smoothed. No spikes at all, just small bumbs.

 

[Skylab]

Quote:

Originally Posted by Bullseye /img/forum/go_quote.gif ^Fletcher and Munson curves might be the ones they are using.

Interesting proposition, and in the bass that would make sense. If the HD800 bass was that flat in reality, it would be very, very lean sounding. But if they have "subtracted" Fletcher-Munson, that would make sense...in the bass. OTOH, if that is in fact what they did, then the treble would be very hot, in reality.

 

[Bullseye]

^Well skylab, Fletcher and munson curves were not studied/built to be applied to lower frequencies. They were studied/built to tame frequencies between 2000Hz-6000Hz (approx, that means +/-, some models state between 600Hz to 4000Hz, etc...) that are related to our human hearing abilities that are focused on those frequencies because they are were human voice is located.

Fletcher and Munson curves lower those frequencies by some decibels in order to hear the full spectrum in what is called "equal loudness" so that due to our hearing abilities no frequency stands over another, (not even the ones our ears are designed to be heard in more detail).

EDIT: The paragraphs above are a rough explanation, but on the article I posted above it is explained in depth. You can also search for more articles related to equal loudness, etc...

EDIT2: In these curve you can see from approx 2k to 6 k how those are tamed:

However these results don't mean they are using equal loudness curves, as I don't know exactly how they measured the headphones, if they planned using those curves or not, ... I am just speculating. Then there is also the question why they only put frequencies till 12k instead of 16k-20k, ...

 

[Skylab]

Quote:

Originally Posted by Bullseye /img/forum/go_quote.gif ^Well skylab, Fletcher and munson curves were not studied/built to be applied to lower frequencies. They were studied/built to tame frequencies between 2000Hz-6000Hz (approx, that means +/-, some models state between 600Hz to 4000Hz, etc...) that are related to our human hearing abilities that are focused on those frequencies because they are were human voice is located. Fletcher and Munson curves lower those frequencies by some decibels in order to hear the full spectrum in what is called "equal loudness" so that due to our hearing abilities no frequency stands over another, (not even the ones our ears are designed to be heard in more detail). EDIT: The paragraphs above are a rough explanation, but on the article I posted above it is explained in depth. You can also search for more articles related to equal loudness, etc... However these results don't mean they are using equal loudness curves, as I don't know exactly how they measured the headphones, if they planned using those curves or not, ... I am just speculating. Then there is also the question why they only put frequencies till 12k instead of 16k-20k, ...

That isn't really the right way to look at it. They were ABSOLUTELY looking at lower frequencies as well - you cannot examine one without the other. The whole point is that the ear is MOST sensitive in the midrange , and how that impacts our perception of frequencies higher and lower. EQ-ing frequencies in the 2-6 K range is no different that EQing the ones outside that range in the other direction

The graphs clearly show that.

 

[Bullseye]

Quote:

Originally Posted by Skylab /img/forum/go_quote.gif That isn't really the right way to look at it. They were ABSOLUTELY looking at lower frequencies! The whole point is that the ear is MOST sensitive in the 1-4K , and how that impacts our perception of frequencies higher and lower. EQ-ing frequencies in the 2-6 K range is no different that EQing the ones outside that range in the other direction The graphs clearly show that.

The human ear in conjunction with the brain is focused on that frequency range you say -as I stated before-. Our perception on sound varies also with the level we are listening to it. Depending on how high we are listening to it the frequency curve has to be changed in order to hear all frequencies on the "same level".

Just check these curves. And relate that to what you said.

Lets not start a discussion over these. I gave people enough information to read about and to think about. Plus this is not the correct forum to discuss about science

 

[Skylab]

Actually that curve clearly shows exactly what I am saying, which is that bass and upper treble frequencies must be louder than the midrange to be seem to a human to be at the same leve as the midrange, and how much louder depends on how loud the sounds are to begin with. This is a function not just of the midrange, but of the whole spectrum.

Maybe we are saying the same thing, just differently

 

[k3oxkjo]

Quote:

Originally Posted by Skylab /img/forum/go_quote.gif I am a 15+ year Stereophile subscriber and am a BIG fan of Keith Howard's articles. I read that article the day I got that issue. I thoroughly understand the difficulties in measuring headphones. I worked for a major microphone manufacturer for 8 years, and I have taught a college class on audiology. I GET IT I understand that one cannot rely on one set of measurements, especially without knowing all the details about how they were taken. But when multiple different sets of measurements seem to show some similar trends, AND the listening you recommended (which I obviously did as well) seem to also correlate, well...I have no issue calling that "evidence" Your point about whether the peak is an "anomaly", or "optimal" is an interesting one, and of course, as you have also wisely pointed out before, that will only be in the ear of the listener, at the end of the day.

Sorry I didn't make it explicit Skylab, I really wasn't addressing the suggestion to read the article to you, I heavily suspected you had from the quality and quantity of your postings historically. It was really for folks who hadn't, as evidenced by subsequent postings.

So this is why I think Sennheiser is insane to release the curves:

The differences we see in the curves between different HD-800 SN's are really pretty small for transducers (Skylab has worked for a transducer manufacturer, perhaps he can weigh in on this). And the assessment of the curves as to their subjective effect is no trivial matter. But people will run with this to try and "explain" what they hear (and complain if someone else's pair measures "better"). I will say that, as much as I respect the work Stereophile has done on measurements, whenever I see John Atkinson in the measurements section say something like; the dip at 4 Khz presumedly explains the reviewer's feeling that the speaker had a bit of a hooded sound, I think, How do you know that's true? Yes, I know he has quite a lot of experience, but really?

Further, any standard for measurement Sennheiser chose would have been second guessed, when really there is no universally accepted standard to start with!

While the raw data is objective, the decision of how to EQ the raw data and subsequent assessment of the EQ'ed curves is subjective. So I guess the point is that the curves, as interesting as they are to see, are not at all definitive.

All of this is not to argue for or against the HD-800 sound quality. At this point, all the listening observations (listening observations? What would George Carlin say about this?) and reviews like Skylab's excellent one taken together seem to point to (IMHO) a fairly accurate Gestalt of the HD-800's qualities so that individal listeners can form a reasonable idea as to whether they would be happy with the HD-800 presentaton. Which, to me, is the best thing reviews can do.

 

[cavedave]

These graphs do not suprise me at all. In the review skylab did I tried to tell you guys that if you set up a stereo system and used a anlizer with a pink noise genarator to get a flat frequency you would pretty much hear what Skylab was reporting he heard. When I did this years ago with a speaker set up when it was all dialed in for a flat response the base sounded a little weak and the high end just a little bright. We think we would like a flat response but the truth is it messes with the way we hear things. When all is said and done mabey the HD-800 was made to well. I have found over the years that most of what we really like to listen to for long periods of time is something less then perfect. Like I said what skylab was hearing is no mystery to me and the graphs bear it out. A flat response is great if you can find a way to roll of the highs a little bit then the base comes into focus to. Which is probaly why a lot of the cans we really like have the treble a little rolled off. Even many CD players do the same thing.

 

[linuxworks]

Quote:

Originally Posted by iriverdude /img/forum/go_quote.gif Would like to see resonse <100hz and >12khz

in marketing, its what they DON'T tell you that you have to worry about.

some creative marketing by senn?

seems so.

for over $1k, I'd demand to see the 'full' 20-20k. and then some! sheesh!

 

[Skylab]

Quote:

Originally Posted by k3oxkjo /img/forum/go_quote.gif Sorry I didn't make it explicit Skylab, I really wasn't addressing the suggestion to read the article to you, I heavily suspected you had from the quality and quantity of your postings historically. It was really for folks who hadn't, as evidenced by subsequent postings. So this is why I think Sennheiser is insane to release the curves: The differences we see in the curves between different HD-800 SN's are really pretty small for transducers (Skylab has worked for a transducer manufacturer, perhaps he can weigh in on this). And the assessment of the curves as to their subjective effect is no trivial matter. But people will run with this to try and "explain" what they hear (and complain if someone else's pair measures "better"). I will say that, as much as I respect the work Stereophile has done on measurements, whenever I see John Atkinson in the measurements section say something like; the dip at 4 Khz presumedly explains the reviewer's feeling that the speaker had a bit of a hooded sound, I think, How do you know that's true? Yes, I know he has quite a lot of experience, but really? Further, any standard for measurement Sennheiser chose would have been second guessed, when really there is no universally accepted standard to start with! While the raw data is objective, the decision of how to EQ the raw data and subsequent assessment of the EQ'ed curves is subjective. So I guess the point is that the curves, as interesting as they are to see, are not at all definitive. All of this is not to argue for or against the HD-800 sound quality. At this point, all the listening observations (listening observations? What would George Carlin say about this?) and reviews like Skylab's excellent one taken together seem to point to (IMHO) a fairly accurate Gestalt of the HD-800's qualities so that individal listeners can form a reasonable idea as to whether they would be happy with the HD-800 presentaton. Which, to me, is the best thing reviews can do.

Another excellent post, k3oxkjo. No argument from me at all on any of your points. Nothing will ever be definitive, except, perhaps, that as time passes, a more clear majority opinion will emerge, which is why I enjoyed and thought there was great value in your own review.

I hope that as more people get HD800's they will both take the time to post their graphs, AND, more importantly, to post their opinions on the sound

 

[k3oxkjo]

Quote:

Originally Posted by Skylab /img/forum/go_quote.gif Actually that curve clearly shows exactly what I am saying, which is that bass and upper treble frequencies must be louder than the midrange to be seem to a human to be at the same leve as the midrange, and how much louder depends on how loud the sounds are to begin with. This is a function not just of the midrange, but of the whole spectrum. Maybe we are saying the same thing, just differently

Owww, Fletcher-Munson curves!

First off, as some may know, the F-M curves are an average, not everyone's hearing works exactly as the published curves (yet another slippery slope), but certainly the trend is real enough. Let's assume for the sake of assessing the F-M curves in isolation, that we have a perfect recording made at a particular spot in the hall and a perfect playback system.

Fortunately, we can mostly ignore the F-M effect (except for one factor I will get to) in this case. The F-M characteristic is sort of the basic EQ of the machine (us) applied equally to any input. So in our perfect case, the live music and the recording, having the same spectral content and level, will be interpreted by our ear-brain system as the same, no F-M compensation needed.

The exception comes because the F-M EQ curve changes "shape" with signal level rather than staying consistant. Peter Walker, the designer of the Quad electrostatic speakers, pointed out that this means there is only one correct playback level where the FR tracks with the F-M characteristic dynamically. This is most apparent in the bass where the lines tend to converge at lower sensitivity levels. So if you listen at a lower level than the original performance, the bass seems to "wash out". But if you apply some sort of F-M EQ for this effect, the bass during louder passages will tend to be too much...

There is so much going on with hearing and playback issues and especially when you are looking at the specific case of headphones, it sometimes seems to me like trying to eat soup with a fork.

And I can't think of a better place to discuss "science" than a thread on FR curves.

 

[Skylab]

Quote:

Originally Posted by k3oxkjo /img/forum/go_quote.gif The exception comes because the F-M EQ curve changes "shape" with signal level rather than staying consistant. Peter Walker, the designer of the Quad electrostatic speakers, pointed out that this means there is only one correct playback level where the FR tracks with the F-M characteristic dynamically. This is most apparent in the bass where the lines tend to converge at lower sensitivity levels. So if you listen at a lower level than the original performance, the bass seems to "wash out". But if you apply some sort of F-M EQ for this effect, the bass during louder passages will tend to be too much...

Ahhhh...another great point. This is the reason I use 80dBA as my calibrated level for reviews is that it minimizes this bass sensitivity issue to the extent that is possible given a safe listening level. If someone listens at 90db consistently, their perspective on the spectral response is likely to be that the bass response is stronger that I do, given my chosen listening level. But I don;t consider 90dB an safe listening level.