[krisjan]

I would appreciate is someone could point me to a link or perhaps share in a nutshell why headphone frequency responses look like they do - I am referring here to the general shape of the chart, and not specifics...

I have been looking at a lot of charts on headroom and comparing phones, and it seems as if in general the shape of the graph remains the same - a gentle hump at around 100-200hz and then sloping steadily downwards to about 4-6khz where there is a series of steep ups and downs...it shocked me the first time - not nearly flat (not sure if the dBr scale is very exagerated) and in the highs such extreme spikes. Is it an inherent problem due to physical constraints of headphones, do manufacturers actually "shape" the curve as such because it sounds good? I am curious and any technical primers into the issue would be much appreciated.

 

[psyllium]

The spikes of the highs at the end of the spectrum is due to the cancellation etc effects inside the ear. Yes, different headphones will do it differently, but a lot of it depends on your own ears.

Have a full read of the How to Interpret Headphone Data - HeadRoom - Right Between Your Ears page to get an idea

 

[Acix]

Flat chart of my Pioneer monitor 10...

 

[b0dhi]

Quote:

Originally Posted by Acix /img/forum/go_quote.gif Flat chart of my Pioneer monitor 10...

Is that measured through a dummy head?

 

[krisjan]

Quote:

Originally Posted by psyllium /img/forum/go_quote.gif The spikes of the highs at the end of the spectrum is due to the cancellation etc effects inside the ear. Yes, different headphones will do it differently, but a lot of it depends on your own ears. Have a full read of the How to Interpret Headphone Data - HeadRoom - Right Between Your Ears page to get an idea

ok - thanks for this.
It seems kind of pointless to make these measurements through a dummy head, if everybody's heads would be different...i'd rather do a type of measurement that shows me whether the actual sound coming out of the speaker gives you a flat frequency response or not (like they do with speakers)...i suppose it's ok if the same dummy head is used in terms of duplicating results for comparison, but how would an ideal frequency curve look in the highs where "You’ll notice all headphone measurements have a lot of jagged ups & downs (peaks & valleys) in the high frequencies; this is normal and mostly due to reflection cancellations in the folds and ridges in the outer part of the ear." just seems like an unnecesary inbetween step the processing of the hrtf and then the guessing in the highs.

 

[Nocturnal310]

from Headroom


How do they measure Headphones data
Quote:

The most important thing in a good headphone measurement is to use a proper head microphone. We use a Head Acoustics Artificial Head Measurement System to perform headphone measurements in compliance with industry standards. Wolfgang, as we call him, has soft outer ears just like you and I. The material is designed to have the same acoustic absorption characteristics as human skin and bone. Wolfie has ear canals, too. At the bottom of each is a calibrated instrumentation microphone. When we place a pair of headphones on this head, the overall acoustic coupling to the microphones is just like it would be on the “average” human. The acoustic response of each head is measured before leaving the factory and calibration curves are sent along with each head. We use these curves to subtract from our measurements to get our reference for “flat”. We also have to make sure everything is very quiet around the head. We have a small anechoic chamber in which we mount the head and other required equipment. The anechoic chamber prevents sound from the headphones from getting out into the room, bouncing off a wall, and then coming back into the headphone and disturbing the measurement. The test signals are generated, and the microphone return signal is analyzed by an Audio Precision System Two Cascade audio analyzer. The headphones are driven by a HeadRoom Max headphone amplifier. All measurements are done with the volume at the headphones at 90dB SPL@1kHz. A Sound Pressure Level of 90dB is a little louder than normal listening volume; headphones should be able to reproduce well at this level. When we perform a headphone measurement we carefully place the headphone on Wolfgang; close the door on the chamber; punch the “go” button on the computer that controls the Audio Precision; the AP executes the measurement script, sequentially performing the tests. When it’s all done, the data is saved to a folder and then uploaded to the website for display. Pretty simple really, once you’ve figured out how to do it---which wasn’t quite so easy; but it was fun.

i could have just posted the link but then why not index the wealth of information on head-fi itself.

 

[krisjan]

ok, let me clarify then...it's fine, our ears do not have a flat frequency response - there's the fletcher-munson, etc. AND it varies depending on the specific physiology of each person too - a very subjective type of perception. Why even try to simulate this system when measuring the output of a sound reproduction system - why not just measure the soundwaves travelling through the air. Speaker responses are not done through a dummy head - and full size headphones are just small speakers positioned outside the ear? It's a very nice set-up as outlined above, and good for them for going through the trouble, but in comparing the data between different full-size headphones i feel one might as well use a reference microphone positioned close to the cup - if you do away with the peaks and valleys in the highs the chart becomes a lot simpler to interpret for the average noob looking at the charts...

 

[Punslayer]

Those peaks and valleys are a part of the headphone driver's response and not the dummy head. The way sound (especially high frequencies) bounces off of your skull and outer ears has a big influence on the way you hear things, but headphones bypass most of these reflections. That being the case headphones are not designed with a flat frequency response in an attempt to correct for these missing interactions. That's part of the reason there's no such thing as a universally perfect headphone.

 

[krisjan]

Quote:

Originally Posted by Punslayer /img/forum/go_quote.gif Those peaks and valleys are a part of the headphone driver's response and not the dummy head. The way sound (especially high frequencies) bounces off of your skull and outer ears has a big influence on the way you hear things, but headphones bypass most of these reflections. That being the case headphones are not designed with a flat frequency response in an attempt to correct for these missing interactions. That's part of the reason there's no such thing as a universally perfect headphone.

but headroom states in their write-up re how to interpret the charts, the following: "You’ll notice all headphone measurements have a lot of jagged ups & downs (peaks & valleys) in the high frequencies; this is normal and mostly due to reflection cancellations in the folds and ridges in the outer part of the ear." meaning the ears of their dummy head of course...

 

[Punslayer]

Actually that's referring to the reflections that get bypassed　by the headphones that the headphone makers try to correct for. The response of the dummy head (which is calibrated by the manufacturer) is adressed in this sentence:

"After that we apply an audio correction curve that removes the head-related transfer function and accurately produces the data for display."

They use the dummy head to provide a realistic　platform for measurement, but the dummie's specific interactions are corrected for in the final graphs.

At least that's my reading of their method. whether there are better ways to measure headphones or not is beyond me

At any rate it's a very different game from measuring speakers.

 

[krisjan]

ok - yes i got that- but my reasoning was that perhaps the valleys and peaks are not compensated for by the hrtf, because why are they still there then? You see maybe the "reflection cancellations" destroy some frequencies to such an extent that the hrtf cannot reliably be used to correct for them because you do not know if the frequencies were there to begin with - hope this makes any sense...i'm guessing. If i am wrong and the hrtf fully compensates, then the only way to explain the valleys and peaks is that they must be present in the sound emitted by the headphone, and thus headphone manufacturer's must have built their phones that way - which does not seem to be what is implied.

 

[Fitz]

Quote:

Originally Posted by krisjan /img/forum/go_quote.gif ok, let me clarify then...it's fine, our ears do not have a flat frequency response - there's the fletcher-munson, etc. AND it varies depending on the specific physiology of each person too - a very subjective type of perception. Why even try to simulate this system when measuring the output of a sound reproduction system - why not just measure the soundwaves travelling through the air. Speaker responses are not done through a dummy head - and full size headphones are just small speakers positioned outside the ear? It's a very nice set-up as outlined above, and good for them for going through the trouble, but in comparing the data between different full-size headphones i feel one might as well use a reference microphone positioned close to the cup - if you do away with the peaks and valleys in the highs the chart becomes a lot simpler to interpret for the average noob looking at the charts...

Except for special designs such as the K1000 or MDR-F1, headphones do not behave like small speakers positioned near your ears. They form a small acoustic chamber outside the ear that has a huge effect on the frequency response. This is an example of measuring a headphone as if it were a speaker as you describe (microphone is 1cm from driver, 1/3 octave smoothing applied to the trace):



Can you get even a remotely close idea of how this headphone sounds? Think about how you would interpret the sound of this headphone from this graph before reading on.













This graph is for a middle production K240 Sextett, and most would describe it as a fairly dark and bass heavy headphone. Is that the impression you got from the graph? I would think the headphone was defective if it actually sounded like it looks in that graph. Small changes in the position of the microphone also result in significant changes of the treble response (especially when moving it off-center), but it certainly doesn't change that dramatically when shifting the headphones around on your head. This is probably the least accurate way you can measure conventional headphones, which would hardly make it easier for an "average noob" to properly interpret the graph.

 

[Punslayer]

Quote:

Originally Posted by krisjan /img/forum/go_quote.gif If i am wrong and the hrtf fully compensates, then the only way to explain the valleys and peaks is that they must be present in the sound emitted by the headphone, and thus headphone manufacturer's must have built their phones that way - which does not seem to be what is implied.

Exactly. Headphone manufacturers build them that way to make them sound more natural. You can even see them in the example posted by Fitz, which shows quite well wht the dummy head is used.

 

[lauinbeta]

i had the same questions myself. these are the answers i got.
http://www.head-fi.org/forums/f133/h...graphs-365769/

 

[krisjan]

to fitz,
ahhh, ok - thanks, that clears up the mystery for me - thanks for the info - is there a primer on how they do the design work then (the manufacturer's) - it must be incredibly complex - you have to consider the materials making up this small chamber, etc. wow...i wonder if they do it through an interactive trial and error - you know making, measuring, adjusting, measuring, etc.