[liamstrain]

 
sure crossovers have their problems, but full range speakers still cannot compete with a tweeter+woofer+crossover design... the peaks/valleys are much worse when you try to use a single driver for all frequencies

 
Generally speaking. Though a well implemented full range (say a well corrected back loaded horn design), can certainly best a even a well implement multi way. 
 
With speakers, you have different issues though. The physical volume of air that must be moved is much greater across the frequency range and volumes required. This puts a heavier technical burden on any one drivers. Multi ways certainly have the advantage there, especially at high spls. 
 
With headphones, these are largely mitigated by the relatively small driver movements required to attain the same frequency response. You don't need the engineering heroics just to make it make the right noises. Tweaking the response, however, gets fun and you start to see all the multi-rigidity structures molded into the single drivers. 

 

[obobskivich]

Liam is right, multi-driver headphones encounter acoustic problems that multi-way speakers have to a lesser extent. The ideal system would use a single full-range driver, and there are many headphones that at least approach that (Sennheiser, Koss, Aude'z'e, HiFiMan, STAX, and perhaps somewhat more contentiously, Ultrasone and Beyerdynamic, all have models that approach that burden); it means less distortion from a crossover + no time alignment issues (beyond phase issues caused by imperfect FR, as xnor pointed out), and will fit in a "standard" size headphone.

Bigshot, I think a "ruler flat" (say 0-inf +/- 0 dB, I know it doesn't exist, but for the sake of discussion) headphone would end up sounding either very n-shaped, or fairly bright; not dull. The bass would be very "thin" and it wouldn't sound "flat" from person to person (because ear shape will inherently force it to not behave like it does in an anechoic test with a coupler). This is based on statements made by Ultrasone engineers and Tyll. I agree on the v-curve point though.

 

[Eisenhower]

Quote:

 
Generally speaking. Though a well implemented full range (say a well corrected back loaded horn design), can certainly best a even a well implement multi way. 
 
With speakers, you have different issues though. The physical volume of air that must be moved is much greater across the frequency range and volumes required. This puts a heavier technical burden on any one drivers. Multi ways certainly have the advantage there, especially at high spls. 
 
With headphones, these are largely mitigated by the relatively small driver movements required to attain the same frequency response. You don't need the engineering heroics just to make it make the right noises. Tweaking the response, however, gets fun and you start to see all the multi-rigidity structures molded into the single drivers. 

 
Using single drivers adds intermodulation distortion, since you trying to reproduce more frequencies at once. 
Multi-driver balanced armature IEMs usually sound better than ones with single drivers.. but obviously most (all?) full sized opened headphones use single drivers for the reasons mentioned.

 

[bigshot]

Bigshot, I think a "ruler flat" (say 0-inf +/- 0 dB, I know it doesn't exist, but for the sake of discussion) headphone would end up sounding either very n-shaped, or fairly bright; not dull..

Flat is within the range of human hearing, which means +/- 1 to 3 dB (depending on the frequency).

I read somewhere there is an EQ adjustment you have to make with headphones to make them sound flat to the ear... Something to do with the drivers being pressed up against the ear. Measuring flat doesn't turn out flat once they're clamped on your head. Someone around here may know what that exact correction is.

But perceived flat eliminates problems with masking and listening fatigue, and gives the most natural sound.

 

[obobskivich]

Flat is within the range of human hearing, which means +/- 1 to 3 dB (depending on the frequency).

Yeah, I got that, I was just saying for the sake of discussion, say we had a perfect transducer. But even with what you're describing, I'm leaning on it sounding bright and gnarly.

I read somewhere there is an EQ adjustment you have to make with headphones to make them sound flat to the ear... Something to do with the drivers being pressed up against the ear. Measuring flat doesn't turn out flat once they're clamped on your head. Someone around here may know what that exact correction is.

There are actually a few different corrections that can be made, Diffuse-Field and Free-Field are the most common and have some degree of standardization. Regarding why it doesn't work out "perfectly" - it's because the acoustic coupling between the driver and the sensory parts of the ear has the outer ear, the seal of the earcup, and the ear canal getting in the way. I know that Ultrasone has researched this problem fairly extensively, and their conclusion is basically that different folks will respond to the same headphone (no matter how well it measures) differently. In their testing of their own headphones, they found that "non-ideal" earshapes more often than not (as in, with statistic confidence) mean users complaining of an over-bright or over-harsh sound, than anything else.

Tyll's "ideal FR" has a roll-off on the top-end, you can see it here: http://www.innerfidelity.com/content/headphone-measurements-explained-square-wave-response-page-2

But perceived flat eliminates problems with masking and listening fatigue, and gives the most natural sound.

Perceived flat is closer to a v-curve if you're trying to compensate for Fletcher-Munson *and* ear acoustics, and it won't be "flat" to all listeners. It may be flat in your model, and for your dummy head, but that's an approximation - not reality.

There are some headphones that are pretty flat overall, but I wouldn't consider any of them absolutely non-fatiguing compared to models that you'd likely say are very fatiguing.

EDIT

Found this ancient post on the topic: http://www.head-fi.org/t/19259/what-is-diffuse-field-equalization#post_215696

If you go find the K1000 thread in Headphones (Full-Size), where one of the principal engineers popped in about a week ago (yeah!), he talks about DF being AKG's goal, at least back in the day.

Also, thanks to DuckDuck:
http://headwize.com/articles/hguide_art.htm#specs
http://www.aes.org/e-lib/browse.cfm?elib=8322
http://north-america.beyerdynamic.com/service/faqs/faq-for-headphones/purchase-advice-for-headphones.html (scroll down to "Localization").

 

[bigshot]

Listening fatigue (unless it's caused by physical things like uncomfortable fit) is usually due to spikes in the upper mids or high frequencies. The spikes can be so narrow, you don't really notice them, but they drill in at volumes above the rest of the music.

 

[obobskivich]

Listening fatigue (unless it's caused by physical things like uncomfortable fit) is usually due to spikes in the upper mids or high frequencies. The spikes can be so narrow, you don't really notice them, but they drill in at volumes above the rest of the music.

Yes I've heard this theory before. The more contemporary argument is that it isn't just random "spikes" in the frequency response, but "ridges" that you will see on CSD. Despite that, there's plenty of people who go along listening to headphones that according to that theory are "unlistenable" or "highly fatiguing" and I have yet to see any conclusive preference study on headphone FR (I keep asking though), so I'm fine to simply accept this as a known unknown and recognize that personal preference will ultimately dictate reality here.

 

[stevenswall]

Quote:

Yes I've heard this theory before. The more contemporary argument is that it isn't just random "spikes" in the frequency response, but "ridges" that you will see on CSD. Despite that, there's plenty of people who go along listening to headphones that according to that theory are "unlistenable" or "highly fatiguing" and I have yet to see any conclusive preference study on headphone FR (I keep asking though), so I'm fine to simply accept this as a known unknown and recognize that personal preference will ultimately dictate reality here.

 
I believe in a theoretically perfect FR curve for the average human ear. That being said, humans may not be close to the average, and the average of two polarized sides doesn't accurately represent both sides.

There's a pertinent quote I'd like to bring up by Sean Olive that goes something like this,  "Sound is not a personal preference, we just think it is." Sean Olive is an engineer for Klipsch.

 

[bigshot]

The flatter I get my system, the better it sounds. My friends agree, but they don't know what I'm doing to get it that good.

 

[anetode]

You're asking a 1-2" transducer to reproduce 20-20k with low distortion and fast decay. Except that you also want to add an HRTF curve in 1khz+ range. And you want to do it without a DSP, using only passive componentry. And you want the whole package to be small enough for people to comfortably wear and carry around. And you want to use materials which are affordable, easy to machine and available in quantities sufficient for tens of thousands of headphones.

Unless you're someone like Sony then you don't have the manufacturing resources and development funding to make it happen. Unfortunately, if you are Sony, you'll notice that people will pretty much buy whatever crap you care to put out because of the strength of your brand, so you shackle your engineers. (All you have to do is let them of the leash every few years to release a low volume statement product and pat yourself on the back)

Face it, physics and economics conspire against us having good headphones.

 

[Posam]

Quote:

You're asking a 1-2" transducer to reproduce 20-20k with low distortion and fast decay. Except that you also want to add an HRTF curve in 1khz+ range. And you want to do it without a DSP, using only passive componentry. And you want the whole package to be small enough for people to comfortably wear and carry around. And you want to use materials which are affordable, easy to machine and available in quantities sufficient for tens of thousands of headphones.
Unless you're someone like Sony then you don't have the manufacturing resources and development funding to make it happen. Unfortunately, if you are Sony, you'll notice that people will pretty much buy whatever crap you care to put out because of the strength of your brand, so you shackle your engineers. (All you have to do is let them of the leash every few years to release a low volume statement product and pat yourself on the back)
Face it, physics and economics conspire against us having good headphones.

Sony as a brand is hurting right now, they should put out a "we made a super headphone", it would get press and help their image. 

 

[obobskivich]

I believe in a theoretically perfect FR curve for the average human ear. That being said, humans may not be close to the average, and the average of two polarized sides doesn't accurately represent both sides.


There's a pertinent quote I'd like to bring up by Sean Olive that goes something like this,  "Sound is not a personal preference, we just think it is." Sean Olive is an engineer for Klipsch.

Dr olive, as usual, is interesting to consider. I think he is one of many good sources of information, although I hadn't heard about him moving to Klipsch - maybe we'll see some good speakers from them again.

I wanted to add that, as with any research involving humans, you cannot have completely objective and dehumanized data; even if you're only responding to physiology you're still responding. A lot of Olive's datasets support this, the merit of Olive's work is that it looks at preference emprically - something that a lot of people don't do. The potential danger is that you take a statisitic average and use it prescriptively for all future subjects, and that doesn't work (it isn't good science).

 

[autumnholy]

Because all theories regarding human derives from samples... and samples... are well, still samples. There may be outliers.

 

[bigshot]

Hairs can be split and resplit, but the truth is good headphones are a LOT closer to flat out of the box than speaker systems. And with a little bit of careful EQing, you can easily correct for their small deviations to a point where human ears perceive it as flat response. That point will provide the most natural and clearest sound possible. From there, you can apply your personal tone control preferences to taste if you really must.

 

[obobskivich]

Hairs can be split and resplit, but the truth is good headphones are a LOT closer to flat out of the box than speaker systems. And with a little bit of careful EQing, you can easily correct for their small deviations to a point where human ears perceive it as flat response. That point will provide the most natural and clearest sound possible. From there, you can apply your personal tone control preferences to taste if you really must.

I agree with the first part, not so much the second - basically I have a problem with the value judgment and dismissal of the outliers. But yes, good headphones are in general more consistent than speakers.