Skip to main content

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: All headphones being equal ? (Read 34672 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

All headphones being equal ?

Reply #25
Or perhaps he didn't quite "get it" the first time around.  He'll look it over again. 

Steve999 already has a copy of the peer-reviewed JAES article that I referenced, that refers to such experiments. However, I suppose that he never got around to reading it, or maybe he just forgot about it.... 

-Chris

All headphones being equal ?

Reply #26
I am NOW READING:

On the Standardization of the Frequency Response of High-Quality Studio Headphones
Theile, Gunther
JAES, Vol. 34, No. 12, December, 1986, Pages 956-969

Interesting that in the paper the much-older-model Beyer DT880 was one of two tested headphones with the "best (that is, the most flat) diffuse field response compared with the unequualzied response of the other [tested] headphones."  The other was the Stax SR Lambda Professional.  I have the 2005 version of the DT880.  Now there's a 2006 version!  The paper shows that the diffuse field response of the old DT880 had the same type of spike between 8 and 10 kilhertz that I have EQ'ed down in my later-version DT880.  The paper confirms that I have gone about it the right way.  Encouraging!  Otherwise, my 2005 DT880's response is quite flat.  (See the similar response curve of the 2006 DT880 at headphone.com.  This newer version has taken down the spike a little, but with some trade-offs, such as a broader band of emphasized high frequencies.)

Anyway, I digress.  Some findings in the paper:

1) "The tone colors produced by headphones of quite different designs were surprisingly similar when electrical equalization was added."

2)  Measurement and reproducability reliability is best for headphones with soft cushions that go around the ear.  This seems the best design for consistent results from person to person.  [I note that the better current Senns, Beyers and AKGs are made this way.]

3) None of the tested models had a flat diffuse field response, though the Stax and Beyer were by far the closest.

4)  The performance of a high-quality studio headphone can be enhanced by equalizing the original diffuse-field response to be flat.

To make a long story short, to respond to the original poster, the paper argues, and supports with objective data, that EQ is an extremely valuable performance-enhancing tool with headphones, and high-quality headphones can be made to sound very similar to one another using EQ. 

Also, one of the key points of the paper was that for headphones a "diffuse field" response was shown to be theoretically and subjectively preferable, as compared to "free-field" equalization.  Diffuse field equaliztion is "defined on the basis of the average value of the transfer functions of the outer ear in different directions."  Further, "free field equalization applied to headphones cannot produce good results."

Fascinating.  Much thanks to WmAx.

Steve999 already has a copy of the peer-reviewed JAES article that I referenced, that refers to such experiments. However, I suppose that he never got around to reading it, or maybe he just forgot about it.... 

-Chris

All headphones being equal ?

Reply #27
Phase response :
minimal phase response means that the amplitude and phase are mathematically linked.
Same amplitude response gives same phase response. So when you equalise a minimum phase transducer with a minimum phase equaliser (every analog one and most digital ones, but not linear phase ones), you get the correct amplitude response and also the correct phase response. I don't know if it is clear enough...

Quote
That is very interesting. Is there a link to the paper?

There are two papers here : http://www.aes.org/publications/preprints/...ints_search.cfm, search for author voinier, each paper is 5$

Paper 1
Transfer functions of headphones were measured on a dummy head in order to find relations with headphones subjective quality. Depending on headphone position on the head, transfer functions may be quite different. A new technique for representative transfer function selection based on impulse response cross-correlation is presented. This technique has been applied to the measurement of twelve different headphones.
Paper 2
In order to evaluate the subjective quality of different headphones, a real time simulation of transfer functions is achieved by adding filters to a reference headphone. Listening with quick switching from headphone to another one becomes possible, then subjective quality evaluation method is improved. The results obtained with this method are compared with those obtained with real headphones.

All headphones being equal ?

Reply #28
when you equalise two different heaphones so to have a similar frequency response, do those headphones exactly sound the same ?


Wmax has given some good pieces of answer :

THD/IMD(non-linear distortions) is probably a non-issue in almost any quality headphone. The human hearing system can not detect the THD spectrum typical of most transducers, unitl is equal to single digit percents,


Two examples of harmonic distortion from headphone.com, read on the first harmonic product in their graphs : Koss PortaPro, 0.07 %, Sennheiser HD-650, 0.01 %. The THD should not be more than twice these numbers.
In order to figure out how much it is, you can add harmonic distortion to your wave files using the freeware Distrorder : http://www.ohl.to/audio-tools-and-thoughts...ins/distrorder/

A very poor headphone probably has very little chance of ever sounding like a quality headphone, due to the probable severe resonances that are beyond reasonable correction(instead of a few localizable ones in a quality device) and the non-linear distortion at low frequencies at medium to high SPL.


It means that in practice, equalizing bad headphones to make them sound like good ones is impossible because the frequency response accidents are too narrow, too big and there are too many.
For example, changing the position or the pressure of the headphones on your ears may move a given resonace a few tens of Hertz away, which will completely untune your equalization, that applies a +10 dB peak at 7500 Hz while the headphones applies a -10 dB dip at 7550 Hz ! You end up with a 20 dB accident instead of a flat response.

The non-linear distortions at low frequencies are just saturation and clipping of the bass. Cheap headphone can't reproduce low frequency at high sound pressure level. Trying to equalize them will just drive them into saturation.

- most headphones are (?) minimun phase tranducers so correcting amplitude also corrects phase


What about resonances between the headphone and the ear canal ? Shouldn't their phase be random ?

- frequency response should not be too different to avoid gross correction (ie high peaks) that may overload your amplifier and/or transducer and bring distortion


Check http://www.headphone.com/technical/product...s/build-a-graph
Comparing two high end headphones, like the Sennheiser HD-600 and the Beyer DT990, we can see that there is a 15 dB difference at 6000 Hz ! The Sennheiser outputs -10 dB while the Beyer outputs +5 dB.

Lots of discussion revolving around frequency response, not enough around phase response.


Distrorder also applies phase response distortion to the signal of your choice, with an all-pass filter.

Phase alterations exist in the digital domain as well I'm afraid.  You can't get around it.


You can have equalisation with or without phase shift in the digital domain. Low pass filters for DACs are without phase shift. Foobar2000 equalizer seems without phase shift too. But software graphic equalizers often have a poor response. Foobar2000 one and Soundforge 4.5 one as well apply the correction to the whole frequency band. For example if you set -10 dB at 1000 Hz in Foobar, you get 0 dB at 995 Hz, -10 dB at 1005 Hz, -10 dB at 1495 Hz, and 0 dB at 1505 Hz ! Soundforge shifts the phase, but in both cases, using them to equalize introduces strong accidents in the frequency response curves, which is not a good thing.

All headphones being equal ?

Reply #29
In order to figure out how much it is, you can add harmonic distortion to your wave files using the freeware Distrorder : http://www.ohl.to/audio-tools-and-thoughts...ins/distrorder/


I glanced at that software, and upon trying it, it appears you can not alter the spectrum distribution of the harmonic distortion. The fixed structure in that software displayed in the spectrograph window is not representative of what occurs with an average transducer. The software in question creates a very broad/flat distribution with very slowly decaying rate of harmonics related to the fundamental, where as an average transducer has a rapid decaying structure, and not nearly as flat a distribution at the first few harmonics as seen in that software. It is critical that this is accurately modeled if accurate results are to be obtained. Adobe Audition has a distortion filter which can be set to replicate different structures, but the filter has what sounds like quantization issues at low level signals, making the filter overall useless. If you can refer to a distortion filter/FX system that is appropriate for psycho-acoustics testing, I would be very thankful. Maybe I missed the options that configure Distorder -- I only used it for a minute -- so correct me if I missed it.

-Chris

All headphones being equal ?

Reply #30

- most headphones are (?) minimun phase tranducers so correcting amplitude also corrects phase


What about resonances between the headphone and the ear canal ? Shouldn't their phase be random ?


This is interesting.  IIRC, when WmAx measured a pair of (2005) DT880s, he found that a narrow spike [which I had located by ear at my old age!] at about 9 khz was due to a resonance between the headphone and the ears.  This is by far (to my ears) the most severe problem with the frequency response of my DT880s.  I have a parametric filter set up on my Behringer DEQ2496 with which I can easily dial this up or down to my heart's content.  I sometimes leave some of it alone, because the sound of the spike is not altogether unpleasant on recordings without much content at 9 khz.  It adds a shimmering quality or sheen.  (The 2006 DT880s would bea different story, judging by the new headphone.com measurements.  The treble emphasis is less in amplitude but much wider.)

Anyway, you are saying that this narrow 9khz spike, if it is due to a resonance between the ears and the headphones, would be "random phase" rather than minimum phase?  Any thoughts on whether this might have audible consequences when applying narrowly tailored EQ to eliminate it?  I will say, that when I dial this spike down, the increase in quality to my ears is dramatic.  WmAx has constructed a physical filter of acoustic foam to place in the headphones that also greatly reduces this spike.  Maybe this would be an even better solution rather than EQ if it is a phase-random phenomenon.

I take it from what WmAx wrote earlier, that it takes a lot for a phase shift to be audible with music.

Incidentally, IIRC, a smaller but wider emphasis of my version of the DT880s that WmAx found in the 5 khz area he determined to be due to the driver and not due to a resonance between the driver and ears.  So I take it this 5 khz empahsis would be a "minimum phase" anomaly, and that correcting it with my EQ would also correct the phase anomaly.  To me, though, this emphasis in the frequncy response is not a problem in terms of enjoying the headphones.  Though I have set up a parametric filter to dial this down if I feel like it.  Aside from that, their frequency response is really quite flat, as I understand it.

All headphones being equal ?

Reply #31
Quote
you can not alter the spectrum distribution of the harmonic distortion. The fixed structure in that software displayed in the spectrograph window is not representative of what occurs with an average transducer

Vmax, you're right.
I commited this software : I just wanted to have a possibility to add various linear and non- linear distortions. It would be quite complicated for the basic user to also change the distortion spectrum. The actual distortion is more like an amplifier saturation than a transducer harmonic distortion. But I wanted to have something easier to detect. Anyway, I have to think about having a more "soft" distortion with maybe less amplitude of the higher harmonics. Thanks for all suggestions.

All headphones being equal ?

Reply #32
about phase response. ive listened to testtracks with strange phase. its very hard to hear phasevariations and i feel its of little importance compared to everything else, distortion frequency etc.

All headphones being equal ?

Reply #33
Now it has been four years since this discussion and headphone manufacturers have come up with a whole phalanx of new, expensive models. Oddly enough, after Sennheiser released their much acclaimed HD800, most of the competitors suddently proclaimed similar technological quantum leaps which allegedly justified putting out models five times as expensive as their previous, long-year flagships.

Before that, the market for headphones had been relatively reasonable. The prices for high-end models where usually around $200 or $300 (electrostatics excluded) - now we are experiencing a bloat in prices and fancy new marketing terms like beyerdynamics "Tesla technology". So the question is: is there any real advencement behind all of this, or is the headphone market just discovering the audiophile segment?

On the other hand, there have been appearing incredible cheap headphones from far-east manufacturers, like the notorious Superlux, which cost about $40 and compare favorably with established top-tier models like the DT880.


All headphones being equal ?

Reply #35
A while ago I posted those links in a german hifi-forum. You can't image the storm of protest I was provoking with that. It ended up whin an overall agreement that this study just prooves that science got it all wrong

I'm just wondering if there could be any justification at all to those new models like HD800, T1 etc.

All headphones being equal ?

Reply #36
A while ago I posted those links in a german hifi-forum. You can't image the storm of protest I was provoking with that. It ended up whin an overall agreement that this study just prooves that science got it all wrong

I'm just wondering if there could be any justification at all to those new models like HD800, T1 etc.



Call me a cynic, but I suspect that headphone price increases are based on an expanion of the market for headphones by people who use portable digital players.

All headphones being equal ?

Reply #37
A while ago I posted those links in a german hifi-forum. You can't image the storm of protest I was provoking with that. It ended up whin an overall agreement that this study just prooves that science got it all wrong

I'm just wondering if there could be any justification at all to those new models like HD800, T1 etc.
Is it really that easy? That would explain a lot. So... let´s try if I got this right: I could simulate the frequency response of a HD-800 with my HD-600 with the result being almost the same sound of the HD-800? I don´t want to protest - I am baffled. Because every expensive headphone would be snake oil then, wouldn´t it?


A while ago I posted those links in a german hifi-forum. You can't image the storm of protest I was provoking with that. It ended up whin an overall agreement that this study just prooves that science got it all wrong

I'm just wondering if there could be any justification at all to those new models like HD800, T1 etc.



Call me a cynic, but I suspect that headphone price increases are based on an expanion of the market for headphones by people who use portable digital players.

Well, sometimes you can be cynical but in this case I´d like to agree with you    Just look at all the headphones that came out painted in a fancy white colour finish in recent years... Also, there now is a market for expensive headphone-only amps. I own one of these myself - and for several months now I have been asking myself if that was really necessary...
marlene-d.blogspot.com

All headphones being equal ?

Reply #38
I think this is a classic system control task. You have a reference system with transfer function (z-transform) of F1(z), this is usually measured using unit impulse input signal and sampling the system's response. Then you have another system with a non-ideal transfer function of F2(z) which you wish to have F1(z) instead. If everything was linear, you could just put the following filter on the input of the non-ideal system Fcorr(z) = F1(z)/F2(z).
Easier said than done, though. Fcorr(z) might not be causal (fixed by introducing artificial delay) or it might produce out-of-range output (possibly damaging the system or hitting full scale limit and not working properly). What about sampling frequency, number of Fx(z) taps (samples) etc...

You could also do that in analogue domain using Laplace transfer functions Fx(s) but then Fcorr(s) might not even be constructible using real world components (C/L/R/opamp circuits). Even if it was, it would probably damage the emulating headphones if they were "too different" from the emulated ones (insane amplitude spikes on Fcorr(s) output required to compensate the difference between the two headphones).

It's been years since I had the course at the university so I might be wrong.

All headphones being equal ?

Reply #39
When commercial records are recorded and processed using one ore more feedback paths (e.g. genelec near-fields), for optimal subjective rating I would expect that similar playback response would be a good starting-point?

I dont worry that much about phase-response. It looks bad on the screen, but does not usually sound bad, so who cares?

I think that Floyd Tooles book summarize it pretty good by saying something like the frequency-amplitude response is the single most important feature of any loudspeaker (or was it hifi system), and most other aspects are easily buried in that "noise".