Head-Fi.org › Forums › Equipment Forums › Sound Science › Why isn't flat on the compensation curve flat?
New Posts  All Forums:Forum Nav:

Why isn't flat on the compensation curve flat?

post #1 of 23
Thread Starter 

So. 

 

The flattest headphone I have, both in listening and graphs, is the V6. I've definitely heard flatter, and a lot of headphones measure flatter. The issue is that headphones that sound flatter, don't really measure flatter. 

 

The V6 is bright. It is the brightest "neutral"ish studio cans I've heard. So, I would expect that the graph of the treble would be tilted up. But it isn't, the treble response on graphs remains between 0 and -3db (inaudible difference I think) of the target line from 2kHz to 5kHz, dips a bit and really starts rolling off past 10kHz. It remains either at or below the 0 level on the frequency response graphs throughout the entire treble, why does it sound treble happy? 

 

I always assumed that my ears just were more treble sensitive than most. The graph says its flat, so I have to remember for myself personally, that flat on the graph means more than neutral level of treble and the V6's are called a bright headphone since most other headphones reduce treble even more so than the V6 so they can sound like its coming from speakers or something that gets treble attenuated by air. Yay. But apparently that isn't true. 

 

Apparently zero on the graph is lying, and zero on the graph isn't really flat, and my ears aren't weird. 

 

So why doesn't a headphone that follows the zero line well into the treble sound flat? 

post #2 of 23

Check out the section on "What is diffuse-field equalisation?"

http://north-america.beyerdynamic.com/service/faqs.html

 

Also, halcyon's post:

http://www.head-fi.org/t/19259/what-is-diffuse-field-equalization

post #3 of 23
Thread Starter 

I get the whole diffuse field equalization thing (maybe), but still, shouldn't zero on a diffuse field compensation curve sound somewhat like a flat loud speaker in diffuse field? Are supposedly flat loudspeakers in a diffuse field environment really that bright? I'm missing something big... 

 

People compensate the headphone measurements. Why don't they compensate them in a way that the zero line represents neutral sound? Or are we just too wildly different in hearing past 2kHz to do that? 


Edited by ThinkAwesome - 6/27/13 at 9:25pm
post #4 of 23
Can you show me the graph in question?
post #5 of 23
Thread Starter 

I was looking at InnerFidelity's Sony V6 graph typing this up, but it is reflected on most graphs of the V6. Though it seems that Tyll actually uses Independent of Direction rather than Diffuse Field. 

 

But basically, if you look at it, the frequency response line ignoring the roll off on both ends remains fairly flat throughout. You see the small dip in the mids, and you hear it, the small hump in the mid bass, and you hear it. You see it remains fairly close to the zero line up until around 4 or 5 kHz, then dips down, peaks at 10kHz and rolls off. If you didn't know that flat on the graph on anything past 2kHz is likely to not be neutral for the majority of people, and if you didn't have graphs of other more neutral sounding heapdhones to compared, you wouldn't expect the brightness you hear in the V6. 

 

Why is there no compensation method that follows what a neutral headphone sounds like? I get that people's ears (and head and torso) are different, but I think that flat treble on a graph consistently correlates with emphasized treble in listening for pretty much everyone. From what I've seen on graphs of headphones I heard, if a headphone is ruler flat on the frequency response graph from pretty much any website I've seen, it would be a bright headphone. 

post #6 of 23
It has to do partly with the diffuse field compensation curve itself being too bright and partly to do with mastering studios making recordings too bright to sound good on bad stereos and takr part in the loudness war at the same time. (since the same digital full range swing on a recording sounds much louder when it is a full swing of treble than of bass)
post #7 of 23

The V6 has quite a peak at 5KHz - 7KHz, I seem to remember. That could tilt what you hear towards the treble. Also, It doesn't really go that low in the bass before it rapidly falls away.

 

I have a modded Fostex t40 which measures flat from quite low, to reasonably high with no real peaks. Funnily enough, it sounds nothing like a V6. It's way smoother and nowhere near as strident so it does make you wonder - what exactly does 'flat' sound like?

 

To get a smooth sound from a flat headphone, it mustn't have peaks or any what is called 'ringing'. (Where frequencies continue after the sound has stopped) The Sony also has strong ringing as well at about 4KHz, I think so that would make it seem bright too.

 

It's not just a question of the FR graph and there are many things that come together to give you the sound that you hear.

post #8 of 23

For a complete normalization of headphone response so that a "flat" headphone shows a flat FR curve, there would have to be a standardized target curve, essentially, a curve that accurately represents what "flat" is for headphones.  There is some disagreement in the industry of what that should be, so at this time there is no standard for normalization.  That's why if you look at compensated measurements from different sources, they end up looking different.  To make matters worse, the target curve would have to be different for each general class of headphone, full, on-ear, and IEM, then within that, there might have to be adjustments for full seal vs partial, vs non-seal.  

 

Standardizing a headphone target curve is not an easy thing to do.

post #9 of 23

Diffuse field is too bright because it does not take into account room effects. If you put an omnidirectional microphone at the listening position in a diffuse field, the microphone would show a flat response. In a typical listening room, the microphone would not show a flat response, but one that is sloped gradually downward. You can see this by looking at target in-room response curves for loudspeaker systems, such as the small-room x curve, Møller room curve, and others. Mead Killion of Etymotic addresses this in these presentation slides (page 30).

 

Technically, using target in-room responses is not quite accurate either because human hearing is not omnidirectional, nor is a typical listening environment exactly a diffuse field. Olive and Welti propose a new target curve based on the eardrum response of a loudspeaker system in their latest paper.

post #10 of 23

I think the post I just completed at InnerFidelity addresses this subject directly:

 

http://www.innerfidelity.com/content/harman-researchers-make-important-headway-understanding-headphone-response

post #11 of 23

Great article Tyll!  

 

Dinner with Olive and Toole?  Do I guess you don't remember what you ate....

 

I love this: "Dontcha just love it when common sense prevails."

post #12 of 23
Quote:
Originally Posted by Tyll Hertsens View Post

I think the post I just completed at InnerFidelity addresses this subject directly:

 

http://www.innerfidelity.com/content/harman-researchers-make-important-headway-understanding-headphone-response

Wow, an extremely engrossing article. Out of curiosity, which headphone came out on top, in both the virtual test and the real test? Did the preferences correlate with price and reputation or not?

post #13 of 23
Thread Starter 

Excellent read. 

 

As for the test. I don't think price/reputation should have had an effect, it was double blind, so they should have had no clue what each of the headphones were (except the heavy LCD2 possibly). 

post #14 of 23
Quote:
Originally Posted by jaddie View Post

Great article Tyll!  

Dinner with Olive and Toole?  Do I guess you don't remember what you ate....

I love this: "Dontcha just love it when common sense prevails."

Simulating LCD2s with a K518... Now that's not what I call common sense! If the simulated phones came out on top all the commonsense audiophiles that say you can't polish a turd with EQ would turn into a lynch mob at the door of Olive-Welti! As it stands, for all we know the simulated phones may have gotten lower ratings simply because the subjects knew something was up when they could change the sound of the phones completely at the touch of a button rolleyes.gif

If just simulating phones could turn in such close results, it's no wonder that I could make cheap phones sound better than expensive ones to me by tweaking the frequency response extensively to fit my preference instead of just blindly following the expensive phones rolleyes.gif It would be like the K518 tuned to fit the Olive-Welti target vs the stock LCD2s...
post #15 of 23
Quote:
Originally Posted by ThinkAwesome View Post

Excellent read. 

 

As for the test. I don't think price/reputation should have had an effect, it was double blind, so they should have had no clue what each of the headphones were (except the heavy LCD2 possibly). 

As in, if the results of the test correlated with the price/reputation of the headphones in question. 

New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Sound Science
Head-Fi.org › Forums › Equipment Forums › Sound Science › Why isn't flat on the compensation curve flat?