[ChildOfInsanity]

Hello,
 
so with todays digital technologies, why aren't all the headphones perfect?
I mean, if you can precisely measure the output of the driver and its frequency and phase responses, why don't manufactures employ some kind of digital preamping or equalizer which would completely eliminate those imperfection? Just equalize the input for the drivers so in the end, they play everything just perfectly flat.
I know there exists calibrating microphones for eliminating room impact on sound from speakers...
 
This thought bothers me for quite a while and I wonder if there are some technological or physical barriers to do this.
 
Any ideas?

 

[RRod]

  Hello,
 
so with todays digital technologies, why aren't all the headphones perfect?
I mean, if you can precisely measure the output of the driver and its frequency and phase responses, why don't manufactures employ some kind of digital preamping or equalizer which would completely eliminate those imperfection? Just equalize the input for the drivers so in the end, they play everything just perfectly flat.
I know there exists calibrating microphones for eliminating room impact on sound from speakers...
 
This thought bothers me for quite a while and I wonder if there are some technological or physical barriers to do this.
 
Any ideas?

 
As long as it's possible to prefer the sound of a particular speaker+room combo to another, then it will be reasonable to release headphones with various linear characteristics. You could of course always aim exactly for some averaged curve like diffuse field or the Harman curve, but then some company will come out with a can that has better "soundstage" or "slam" and then it's back to the races.

 

[ProtegeManiac]

Originally Posted by ChildOfInsanity /img/forum/go_quote.gif

 

so with todays digital technologies, why aren't all the headphones perfect?

----

This thought bothers me for quite a while and I wonder if there are some technological or physical barriers to do this.

 

Because the problem is in analogue, and materials. They don't design the drivers skewed, they just choose what version of skewed the engineers and the bean counters can live with. There is no single driver with a perfectly flat response from 20hz to 20khz regardless of at what scale (ie how far from it you measure) it has to do that - driver mass, mechanical movement, impedance shifts, and a number of other factors can affect linearity or even cause some amplifiers to distort with it in some way.

 

In the case of multiple specialized drivers, as much as you can use several types and keep them playing only the range where their response is smoothest (again, even then it won't be ruler-flat), adding more drivers doesn't mean you simply end up with an overall smooth response because crossovers don't exactly stop dead at the center frequency - there will be overlaps and phase issues just from the crossover being there (which is why FR driver enthusiasts would rather listen to a single driver doing 50hz to 10khz). Add to that how in speakers you get more phase issues just from having more drivers since they'll each have significant pathlength variations to your ears, which is why Focal's Utopia series are built like that but even then a significantly shorter or taller listener can still encounter problems if they use the same chairs as the average height range as they can put their ears far enough out of the range where drivers' timing is aligned. Properly designed IEMs minimize this by funneling the sound into a tuned tube, kind of like how some exhaust system manufacturers minimize the separate gurgling of a V-engine's pistons by adding an H- or X-pipe after the headers to integrate the sound from both banks.

 

Emphasis on "all" will be apparent later.

 

 

Quote:

Originally Posted by ChildOfInsanity /img/forum/go_quote.gif

 

I mean, if you can precisely measure the output of the driver and its frequency and phase responses, why don't manufactures employ some kind of digital preamping or equalizer which would completely eliminate those imperfection? Just equalize the input for the drivers so in the end, they play everything just perfectly flat.

 

First, that means they have to sell an entire system specifically tailored to work with each other. Not that it hasn't been done, because it has - JHAudio released a flagship IEM that came with its own DSP box that integrates the active crossovers and amplifiers, similar to how in a serious car system you have a DSP that splits the frequencies and then sends them out to each amplifier channel, except in a car the primary reason for doing so really is so that you can apply custom time delays to each of them thereby synchronizing when the sound from the driver's side tweeter, driver's side midwoofer, passenger side tweeter, passenger side midwooger, and subwoofer all reach your ears at the same time like you're sitting in the middle of two speakers with the subwoofer in front and between the speakers instead of in the luggage compartment behind you. B&O also sells similar systems at home and later was contracted by VW to do the same for Lamborghini (it was around that same time that you no longer get Italian supercars with Alpine stereos, which also stopped putting the Diablo's silhouette on its instruction manuals).

 

Basically it'll be like selling a studio monitor, where you have an active crossover applied before the amplifier but with a DSP applying the EQ effect. The problem however is that it isn't that simple primarily because each individual use will have different needs. You can't apply this system design to "all" products because not everybody will use them as such. Take the JHAudio DSP+IEM flagship for example - not everybody runs around with a "stack" of portable equipment like a miniature home system. Even audiophiles serious about quality playback would sometimes prefer to just hook up a JHAudio CIEM to an iPhone for convenience. Similarly for cars not everybody will pay $120,000 for a Jaguar with a B&W system and instead get a $35,000 car and then spend $5,000 on an audio system (plus installation costs and sweat) that has a DSP that can correct for that particular car.

 

Similarly there is at least one integrated home audio system where you get one console (kind of like the systems from the 1960s, not necessarily what a DJ uses) that has the CDP, DSP, and amplifier in it, and this comes with the amplifier. Here's the problem: that ssytem thus far forces you to use CDs, so any other digital source is a no-go, much less a TT. Second, not a lot of people were willing to shell out the kind of money Weinstein (yes, the piano manufacturer) was asking for, despite the fact that at least unlike a headphone system it's not like you'll be moving this system around. Then there's that little bit about maintenance - if the CDP's transport stops reading discs, your entire system is unusable.  The best we have really are the car DSPs and that JHAudio system, but the latter apparently didn't make enough or can't continue to make enough for it to remain profitable in the long run.

 

These aren't the only limitation in terms of marketing however...

 

Quote:

Originally Posted by ChildOfInsanity /img/forum/go_quote.gif

 

I know there exists calibrating microphones for eliminating room impact on sound from speakers...

 

 

...as we can take for example forum reactions to the DSP tuning mic on my ///////ALPINE processor. Basically, it involves using such equipment properly, and second, users' perceptions of what the results would be are kind of skewed. Prior to that Audyssey-based auto calibration what we did, for those who actually go that far for their cars or home listening areas, was use an RTA with a mic positioned where the listener's head would be. You'd think that would be very scientific, but then the Alpine-Audyssey system came out and people were saying it sounds "wrong." People thought that Alpine's lab got its perception of correct sound wrong, but what was actually happening is that during the set-up the auto system was picking up sounds from outside the car and then correcting them with the EQ.

 

We've basically been doing the same thing since the RTA didn't know which is background noise either, but even then the results were somewhat different due to the algorithm on the DSP. At the same time it wasn't really comparable because as much as it can apply a 500pt EQ correction (plus time alignment) it was also correcting for its very different input: it was an integration processor that takes a high level input from a stock system and then corrects it, but people are comparing it their CDP+DSP combos that don't have to do that. Basically, even the most complex EQ with a scientific algorithm can actually suffer potential issues if it actually has to apply enough EQ effects, while being forced to choose what to correct means less EQ issues but at the same time has also contributed to what people perceive as "natural" not unlike why vinyl enthusiasts think that analogue is that pure. Driver design limitations will also mean that some will need more EQ and therefore potentially just run into the problems that EQ can cause. Basically, EQ nor even DSP are not magical panacea - continuing driver design development is still the end goal. Even in cars despite the need for time alignment to simulate a central seating position the most complex DSP from Alpine or Pioneer or Sony can't correct improperly installed drivers that lead to a lot of cancellations and unwanted reflections.

 

 

Even perception doesn't stop there. As much as the example about the car systems is of people who try to be scientific about making a neutral system (and believe me, I've heard many different cars using very different systems whose sound were indistinguishable if I somehow managed to get into the cars blindfolded) there are people who prefer "lush, creamy, syrupy, audiophile snake oil" sound. While manufacturers aren't doing this end users have access to technology that can - like instead of limiting EQ settings there actually are apps that have custom EQ profiles to flatten every headphone as much as possible, like Accudio. A couple of problems: very few actually like the "flat" EQ correction, despite the fact that it takes the guesswork and sweat out of having to set the EQ yourself on a program that given the accessible interface or other limitations can't apply as many corrections as needed. Add to that how much the Accudio interface sucked, especially if you're using an iPad as a music server (like I do) since it was made for the iPhone.

 

Add to that the final hurdle: every individual listener can have variations in their own hearing that a system that measures flat can sound unnatural in two different ways to two musicians. This would again mean more dependence on EQ, and even if you can forego or have technology that can minimize the problems EQ can introduce will just mean much higher costs as you need to have your ENT/Audiologist coordinate with whoever is selling equipment to you. I'm not saying it can't be done at all, just not to "all" as you think it can be done to.

 

As for me, I'll just keep on using Neutron with the headphones with the smoothest response possible* and then apply tiny EQ changes to smoothen the response on my bedside rig. On the go it'll be a similar thing with an IEM. In my car however a lot of other considerations are necessary, like avoiding tweeters that are too large to fit properly (that said I'm using home audio Vifa tweets with 3in wide mounting plates and 2.5in wide magnets) or one with a dispersion pattern that is too narrow or too wide (these will be known as the ones that are very hard to aim and center). 
 
 
 
*Some headphones are nearly ruler flat up to 1khz, but above that can have a hollowed out midrange smack next to a sudden treble spike, in which case it's easier to apply a wideband cut to the bass plateau of a good dynamic driver as well as disparate cuts on its far apart treble peaks rather than correct the aforementioned issue on the midrange and treble of some orthodynamic drivers

 

[castleofargh]

we'll get to it someday, but out of a few brands who really know what they are doing, many brands wouldn't have a job to go back to, if ruler flat and low distortion was a known fixed value and consensus of quality(not saying it wouldn't be a great thing for the consumer).
then some who are only into headphones would need to get up to date in the DAC/amp/DSP department.
 
and of course there are all the oblivious audiophiles who decided with no understanding whatsoever that all EQ or DSP are destroying the "natural" sound, when nothing is natural in a record, or listening to it using a headphone. changing the mind of the consumer is just as important as changing the product, because the brand will sell what we ask for, even if it's a really stupid product.
 
the last problem is that there isn't one neutral that fits us all perfectly(but we sure could get closer than what we have right now in the headphone business). 1 neutral works for speakers because they're similar to the real life way of listening to sound, so any change from our body, any compensation from our brain, will apply to speakers the same way they apply to everyday life. but with headphones and IEMs, we bypass some of those "custom changes", and when it's not binaural music, there really isn't much that will sound "real".
so the job might be a little more complicated than just making a system that aims toward a given signature. but we sure could all benefit from embracing the potential of digital processing as a mean to improve audio. I'm very much a believer of this.

 

[Wembely]

I don't think you are that far off on your question at all.  In fact there are some very good Apps (for Apple devices anyways) that do EXACTLY what you are talking about.  I used them all the time and in fact typically will not buy or listen to a Headphone or IEM that does not have a digital compensation curve for it. The two I use is Accudio and Audessey. They are full music players that include DSP compensation as you original question posits.
 
They both take detailed measurements of IEM and headphones (and in some cases speakers)  and create compensation files.  Accudio shows before and after graphs so that you can see how close to "flat" a particular IEM or headphone can get.  They rate the DSP compensated result on a scale from 1 to 5 in .5 increments.  I find anything from about a 3.5 up is pretty good (depending on how it gets its grade).  They strive for a flat response based on a diffuse curve.  I happen to think they have picked curve that really works well and the compensated sounds seems just about right.  If you don't think its perfect - it also includes a five band equalizer of sorts  that you can use to modify the standard compensation curve - as well as simulation modes tat in theory can make one set of cans or IEMs sound like some of the best IEM in the industry (though I have some doubts as to how close they actually get).  they also allow you to play FLAC files it your hearing is good enough to tell the difference - I can not so it not worth the storage space to me.
 
Audessey is more of a black box.  They do not show before or after graphs or results of their compensation technology.  Some of DSP compensations they have come up with are really amazing and sounds flat and spot on to me.  Other times it seems like they add in extra bass because that's what they figure someone who buys a particular set of cans wants.  They do not have a ranking for how well the compensated result comes out or how close to flat it is (again flat does not always seem to be their goal). But for some sets, Audessey does a really good job on IEM or cans  that the Accudio app does not offer curves for (i.e. Harman Kardon CL and BT and the UE 500 IEM are amazing as are the Sony EX58v).
 
When I have a choice - I tend to default to the Accudio app since they sound more "right to me", though some times it's really close in terms of tuning between the two.  Other times, the extra bass in Audessey hurts clarity a bit  (but they also have a bass and treble control to fine tune their curves as well). 
 
Audessey charges for each compensation curve now, Accudio does not - you get them all when you buy the app.  They both add new IEMs and cans on a regular basis.
 
These two apps are why I stick to Apple Phones and DAPs to listen to music since they are not available on Android or Windows platforms (they I know of ). I find I much prefer an accurate and subjectively flat frequency response curve when I listen to music. These Apps can take good IEMs and Cans and make them extraordinary IMHO. 
 
As always your mileage may vary...
 
Hope this helps and good luck!

 

[sonitus mirus]

I want a dummy head with a mold of my ears that has built-in mics and connects to any PC that uses some special audio analysis software.  The goal would be to place the headphones over, on, or in the dummy head's ears, click start on the analysis application that would then sweep through the complete audible frequency range while adjusting the levels so that the headphone transducer outputs are flat in frequency response and matched in volume.  The EQ profiles could then be saved and inserted into audio players.  No headphones would be perfect, but they could be tweaked for the most potential.
 
A generic profile could be created that could be sold and used by others with the same headphones or the rights to use such EQ profiles could be monetized with existing audio players.  Audiophiles would probably gobble-up such a gimmick if it came with a cool name and a glowing 6Moons review.

 

[Wembely]

That's pretty much what I described above. Except they are not "your" ears but a standard ear. To me they get close enough to mine. Good luck

 

[DougD]

  I want a dummy head with a mold of my ears that has built-in mics and connects to any PC that uses some special audio analysis software.  The goal would be to place the headphones over, on, or in the dummy head's ears, click start on the analysis application that would then sweep through the complete audible frequency range while adjusting the levels so that the headphone transducer outputs are flat in frequency response and matched in volume.  The EQ profiles could then be saved and inserted into audio players.  No headphones would be perfect, but they could be tweaked for the most potential.
 
A generic profile could be created that could be sold and used by others with the same headphones or the rights to use such EQ profiles could be monetized with existing audio players.  Audiophiles would probably gobble-up such a gimmick if it came with a cool name and a glowing 6Moons review.

 
It's a great idea. 
 
Sonarworks is doing it already, I am rather surprised that they are not better known here on Head-Fi.
 
link to Sonarworks' website
 
(edit: they have "average" equalization templates for common h/ps, and can do custom eq for specific h/ps that customers send them. They do not (AFAIK) do custom HRT EQ for specific measured customers' unique heads and hearing.)

 

[ChildOfInsanity]

Thank you guys for all the suggestions, I will definitely try the apps for my iPhone.
 
It's real pity that more manufacturers aren't using this kind of technology to make the headphones better. I understand that there is no golden standard of the true natural sound, but it would be no problem to adjust the equalization curve to anybody's liking.
I wonder if some of the bluetooth headphones already use something like this, as they have their own amplifier right? Or the Philips Fidelio with their own DAC and amp.
 
Anyway, the sonarworks solution looks great and the mic is not that expensive. I may try it for both my home stereo and headphones...

 

[Wembely]

You are welcome.  Again your are generally correct in your assertions implied by your question.  if you look at self powered studio monitors, many of the top models (doesn't mean expensive but as this is trickling down rapidly) use DSP processing to correct for phase, frequency and impulse responses.  A couple of high-end manufactures in particular sell DSP enhanced speaker systems in the many kilo-buck price range. 
 
Several Noise Canceling headphones I know of also use this to change the sound profile when the NC is engaged (some however do not improve them for the better IMHO since they tend to think MOAR BASS is always better).  Though I have no evidence one way or another, bluetooth/powered cans could and I would suspect in some cases do use DSPs to tailor the sound profile to a designers goal. Whether the goal is a flat or neutral response is another matter.

 

[jcx]

most stereo or multichannel music is mixed on speakers in rooms, expected to be listened to in rooms over speakers
 
up to 1/2 the acoustic power at the listening position in a room can be reflections/reverberation/resonance/room modes
 
so the sound is influenced heavily by room dimensions, wall/boundary/furnishing absorption/reflection vs frequency, loudspeaker placement and loudspeaker directivity - how the radiated sound varies with angle
 
you can easily find commentary from production engineers having multiple rooms, speakers, some even say they play a mix in their car to hear what the consumer is likely to experience
 
 
so what is "the right" response for a headphone? - good luck with nailing that down, getting broad agreement
 
 
of course even using EQ doesn't seem to fix what EQ could be expected to fix in studio monitoring rooms: http://seanolive.blogspot.com/2009/10/audios-circle-of-confusion.html

 

[phoenixdogfan]

There is a company out there called Ossic which is going to roll out a VR nased headphone technology this January which will calibrate the user's unique HRTF, provide  head tracking and an improvement on 7.1 surround which they call "spherical sound".  Some of the design team worked  for Logitech, so they seem to be very serious about this.  This could be a revolutionary product.

 

[jcx]

Smyth Realizer is how old now?
 
http://www.head-fi.org/t/418401/long-awaited-smyth-svs-realiser-now-available-for-purchase
 
and not exactly the 1st movers in the "virtualization" field

 

[phoenixdogfan]

Very expensive, uses an 8 year old box. Given that it's just a microprocessor running a program, I have to ask why  a software based solution running on a modern pc can't equal or exceed the Realizer for a fraction of the cost. 

 

[DougD]

  most stereo or multichannel music is mixed on speakers in rooms, expected to be listened to in rooms over speakers
 
up to 1/2 the acoustic power at the listening position in a room can be reflections/reverberation/resonance/room modes
 
so the sound is influenced heavily by room dimensions, wall/boundary/furnishing absorption/reflection vs frequency, loudspeaker placement and loudspeaker directivity - how the radiated sound varies with angle
 
you can easily find commentary from production engineers having multiple rooms, speakers, some even say they play a mix in their car to hear what the consumer is likely to experience
 
 
so what is "the right" response for a headphone? - good luck with nailing that down, getting broad agreement
 

 
As consumers (aka people with money who pay for things), we should probably set a new expectation that albums should routinely come with five or more separate mixes on the disk. 
 
(1) regular, i.e. stereo mix
(2) 5.1 mix
(3) headphone mix
(4) FLACs (redbook or better)
(5) automotive mix (also FLAC)
 
A bonus could be a enhanced version for bassheads.
 
DVD and BluRays have the capacity. Shouldn't be a problem from a technical point-of-view. 
 
The "only" issues are marketing decisions by the recording labels.