Head-Fi.org › Forums › Equipment Forums › Sound Science › Skeptico Saloon: An Objectivist Joint
New Posts  All Forums:Forum Nav:

Skeptico Saloon: An Objectivist Joint - Page 41

post #601 of 1002

Thank you bigshot

 

I just finished about a month of research based on planar magnetic transducer used on headphones I did for an engineering class at my university, wasn't too serious but just a rough look at the tech. I was fairly certain that what made the planar magnetics different is their airflow or how energy was transferred to air particles. That extra encasing does seem to be unconventional as it looks like it would limit the airflow.

 

The headbands should not affect sound at all but earpads can. It is why Mr.Speaker can supposedly make their maddogs "sound" like an LCD-2.

 

stock t50rp:   http://www.innerfidelity.com/images/FostexT50RP2011B.pdf 

to maddogs:  http://www.innerfidelity.com/images/MrSpeakersMadDogA.pdf

vs LCD2:       http://www.innerfidelity.com/images/AudezeLCD2SN5312123.pdf

post #602 of 1002

Good grief. The bizarro world alternates between "measurements mean nothing- use your ears" and "ears mean nothing because they aren't as sensitive as machines". I analyzed the frequency response the way I always do, using a tone generator, an equalizer and my ears, and these folks are telling me that my measurements are completely subjective and worthless.

 

They swear that an audibly balanced response is theoretically impossible and I am just making it up. Instead, they show me headphone measurement charts with 10dB boosts at one end and and 20dB dips at the other and say that sounds balanced to them. They go on and on about model heads and correction curves to account for ear canal shapes, but none of that applies because I tuned the response by ear. I think the whole idea of the headphone group is to totally obfuscate so everything becomes confused and everyone is right and everyone is wrong.

 

Sure, my own hearing may rate as +/- 2 or 3dB and I may not hear balanced in the highest half octave of human hearing, but that really doesn't matter. Flat to +/-3dB with a rolloff above 12-14kHz is pretty damn good for a set of cans. And yes, I do know what those numbers sound like!

 

(Shaking my head)

post #603 of 1002
Quote:
Originally Posted by KamijoIsMyHero View Post
 

I just finished about a month of research based on planar magnetic transducer used on headphones I did for an engineering class at my university, wasn't too serious but just a rough look at the tech.

 

Maybe you'll find a clue by looking at these... http://www.bgcorp.com/planar-magnetic.html They are the planar magnetic speakers designed by the guy who designed the PM-1s.

post #604 of 1002
Quote:
Originally Posted by bigshot View Post
 

They go on and on about model heads and correction curves to account for ear canal shapes, but none of that applies because I tuned the response by ear. 

 

No, this part actually matters because you tuned it by ear. The response everybody gets in real life and on speakers at the ear drum depends on their ears (ear canal, head, etc.), and the response you get with headphones at the ear drum isn't affected in the same way by the same anatomy. If you tune headphones to be flat to you, it will be a little off to other people—though much better than the vast majority of headphones, of course. At best, all that can be done is for a headphone to hit a target for the average person, so to speak.

 

Tuning speakers flat by ear, if you do it right, works for everybody. Tuning headphones by ear or by machine or whatever method to get it flat for one target or person gets it right for that target or person. It's not universal. In other words, if you tuned it to a target, that would be a reference point. However, if you tuned it by ear for yourself (for the sake of argument, assuming they did it 100%), nobody would know what the response would exactly be unless you measured the response in your ear canal. It's a valid target but not a reference point anyone else would know.

 

In practice, nobody's hitting the target they want exactly anyway, so...


Edited by mikeaj - 4/6/14 at 7:17am
post #605 of 1002

but then again there is factual objective measurement, and there is the rest.  saying that anything not measured under conditions is wrong is going a bit fast IMO.

if bigshot spent years working with pros, his "by ear" might not be your random new to the hobby "by ear" and thus, not so far away from truth as one might think(even if as aways we would have to decide upon what neutral is).

just like people always put too much contrast and saturation in post treatments of pictures, and would never get a good white balance. when on the other hand, most pro photographers would go for something very close when setting white balance(not counting artistic choices), because they have experience, and spent time seeing how the real thing should look like with perfectly calibrated systems. so it doesn't matter how many green cones they have in the eye, they just learned how it should be. it doesn't give the perfect result, but close to it yup!

 

I'm kind of OK with both sides here, let's not make it a matter of principle. bigshot was just sharing his discovery of some gear thinking it might interests us, he's not going to buy 3000$ power cable tomorrow, and wasn't trying to say that we all should give up on science.

post #606 of 1002
Quote:
Originally Posted by mikeaj View Post

No, this part actually matters because you tuned it by ear. The response everybody gets in real life and on speakers at the ear drum depends on their ears (ear canal, head, etc.), and the response you get with headphones at the ear drum isn't affected in the same way by the same anatomy. If you tune headphones to be flat to you, it will be a little off to other people—

How much off?

This is the absurd thing... The whole point of tuning EQ by ear is to get it ballparked so it sounds good to the typical range of human ears, not to refine the measurements down to inaudible levels that only machines can hear. If your ears are 1dB or 2dB more sensitive than my ears in some frequency range, it's not going to make diddly squat difference. Human ears aren't sensitive enough for absolute precision to matter. When you tune by ear, you're working within the range of perceptual thresholds anyway which have a certain amount of wiggle room. A balanced response doesn't have to be stone flat to a microphone with super audible ability to measure. It just has to be basically flat to within the tolerances of normal human ears. People keep jumping to absolutes claiming if it isn't completely measurably flat to inaudible levels, it isn't flat. That isn't true. Human beings prefer flattish response curves. The response curve doesn't have to be any more precise than human ears are.

Now I can see someone saying, "Your hearing above 15kHz is probably not accurate enough to balance EQ by ear" and I would certainly agree with them. But frequencies in that range don't really matter anyway. And they might say, "You are off by 3dB here and there." And I wouldn't doubt it. But the sound to typical human ears is balanced either way. Close enough for government work.

It's bizarro world when people post purely subjective impressions of headphones with absolutely no baseline and everyone goes "Ah yes! The veil is lifted!" and "The PRAT on these is wonderful!" And then someone does an EQ test by ear and everyone jumps in and says "Wait a minute! You aren't accurate to within a fraction of a dB!" Even if I had jerry rigged a dummy head and used a microphone, the same people would be complaining that I wasn't applying the proper correction curve because everyone has a slightly different reaction to Fletcher Munson, or that variation in ear canal or head shapes would alter the results. That's just counting angels dancing on the heads of pins. Some people just want to quibble for quibbling sake.

The point is, I used one technique to determine the basic response of these cans (tuning to acoustic music) and my buddy used another method (tuning to test tones) and both techniques and both of our sets of ears ended up at the same basic place... a 3 or 4dB boost at 3kHz and 6kHz. Two sets of ears, two different techniques, same basic results. I described exactly how we did it and folks can take it for what it is. Maybe we are off by a coupla few dB here and there. But once the amateur scientists in the crowd get a chance to dig out their wig heads and microphones and test the Oppos, the broad strokes conclusion we came to will probably stand- the Oppos have a remarkably flat and natural frequency response.

This sort of technical absolutism is why I really don't feel totally comfortable with either subjectivists or science nerds in home audio. My approach is to apply scientific principles to achieve great sound, but I don't kid myself... ultimately, I don't have to be any more accurate than my decidedly mortal ears. And they give me enough latitude that I shouldn't need to break a sweat to get good sound.
post #607 of 1002
Quote:
Originally Posted by bigshot View Post


How much off?

 

A few dB on good, consistent headphones. More than that, particularly in bass (due to seal, and everything's position dependent of course), on some others. If it were fractions of a dB, I wouldn't really bring it up other than as a curiosity.

 

On a related note, I know I've seen diffuse field target for different people before, showing what the average gets to be based on the individual responses. I can't remember where, though. I guess it wouldn't be that hard to find.

 

But anyway, see here (pdf is in the link):

http://seanolive.blogspot.com/2013/04/the-relationship-between-perception-and.html

 

vertical divisions are 5 dB

 

That's the measured response in the ear canal for another high-end planar magnetic headphone, the LCD-2, which produces more consistent responses than many other headphones. It shows left and right ear response for four different people, so I guess that's eight curves. That's say a 5 dB spread throughout the lower treble where the ear is most sensitive, so definitely on the order of magnitude of changes you were suggesting. Also, the spread is not consistent throughout the range. For example, one curve has around a 2 dB dip from around 1.5 kHz to 2.5 kHz, while some others show about a 5 dB dip from 1.5 kHz to 2.5 kHz, and these figures look pretty smoothed too.

 

The top octave stuff is a crapshoot, but it doesn't matter much anyway, as you say. And even if the response is off in the 3 dB range here and there for some people maybe, that's still going to sound more or less right, at least way better than most headphones. 

 

You can look through the results for the other headphones in the link.


Edited by mikeaj - 4/6/14 at 4:37pm
post #608 of 1002
Well, based on my own thresholds (determined by playing with a sound editing program) I can discern 1dB with tones and 3dB with music. So that graph you posted there would sound pretty much flat to me with any of those colored lines. Can't tell what the horizontal scale is, but I'm assuming that dip low is in the last octave or so. A little more variation at the ends is fine because human hearing isn't as sensitive out at those extremes. Naturally, the lines diverge most in the extreme upper frequencies that don't even really exist in music.

My ears would be pretty happy with the headphones that graph represents. I imagine precise measuring of the Oppos isn't going to look too much different. My measurements didn't stray much more than 3dB in either direction... add a 2 or 3 dB variability because of my own peculiar ears and it might look exactly like that.
post #609 of 1002

I'm envious of your level-headedness bigshot, taking part in that conversation would really make my heart pump.

I don't see any major fault in the methodology applied here.

The crucial part is a common real world sound reference, doesn't much matter if it's from a chamber orchestra, a noisy high-way or restaurant chatter. As long as you adjust the sound reproduced by the transducers to match the real world sound as closely as possible, the response particularities of mine, yours or anyone elses ears doesn't matter one bit. It's beautifully simple, maybe too simple for people to accept it?


Edited by limpidglitch - 4/7/14 at 6:24am
post #610 of 1002

Maybe I'm just badly mistaken or explaining it poorly? I'm hardly an expert in HRTFs.

 

There are two ways you might say the real-world response varies from person to person:

1. factors attributing to body, head, ear, ear canal, etc. shape, resulting in different amplitudes and phases of sounds at the eardrum (see: concha gain, ear canal gain, pinna gain)

2. factors from hearing loss

 

(1) is affected differently when listening to sounds generated from out in space as opposed to originating from right next to your ears (headphones). This is because of the angle of incidence onto the ears and so on and how sound is funneled to the eardrum based on that. And on headphones it depends on positioning and seal too, in a nontrivial way. We'd expect (2) to be consistent no matter the source of the sound (I think?), so that isn't really important for this discussion unless a person has more acute loss in say the midrange.

 

So the difference in response between real life / speakers vs. a particular headphone varies by the person, both at the eardrum and perceptually. This is the key point. If a headphone is calibrated to be perfectly flat (or sounding like any reference you want) for one person, it will not produce the same response for another, generally by some amount measured in whole numbers dB at important frequencies. That said, again, most headphones are off for everybody by more than a few dB at important ranges. Maybe this isn't a huge deal as driver matching and differences in response based on small positioning changes of the headphone on the head make about just as much of a difference or close to it. And yet, some people spend so much time worrying about much smaller effects from amps and DACs and whatnot.

 

I wasn't taking offense at the methodology, just pointing out and quantifying this relevant caveat so the discussion can be a little more precise.

 

 

Quote:
Originally Posted by bigshot View Post

Can't tell what the horizontal scale is, but I'm assuming that dip low is in the last octave or so. A little more variation at the ends is fine because human hearing isn't as sensitive out at those extremes. Naturally, the lines diverge most in the extreme upper frequencies that don't even really exist in music.
 

Yes, it's as you say. Like just about every audio graph with frequency on a log scale, those markers at the bottom are at 10^2, 10^3, and 10^4 Hz.


Edited by mikeaj - 4/7/14 at 8:23am
post #611 of 1002
Quote:
Originally Posted by mikeaj View Post
 

Maybe I'm just badly mistaken or explaining it poorly? I'm hardly an expert in HRTFs.

 

There are two ways you might say the real-world response varies from person to person:

1. factors attributing to body, head, ear, ear canal, etc. shape, resulting in different amplitudes and phases of sounds at the eardrum (see: concha gain, ear canal gain, pinna gain)

2. factors from hearing loss

 

(1) is affected differently when listening to sounds generated from out in space as opposed to originating from right next to your ears (headphones). This is because of the angle of incidence onto the ears and so on and how sound is funneled to the eardrum based on that. And on headphones it depends on positioning and seal too, in a nontrivial way. We'd expect (2) to be consistent no matter the source of the sound (I think?), so that isn't really important for this discussion unless a person has more acute loss in say the midrange.

 

So the difference in response between real life / speakers vs. a particular headphone varies by the person, both at the eardrum and perceptually. This is the key point. If a headphone is calibrated to be perfectly flat (or sounding like any reference you want) for one person, it will not produce the same response for another, generally by some amount measured in whole numbers dB at important frequencies. That said, again, most headphones are off for everybody by more than a few dB at important ranges. Maybe this isn't a huge deal as driver matching and differences in response based on small positioning changes of the headphone on the head make about just as much of a difference or close to it. And yet, some people spend so much time worrying about much smaller effects from amps and DACs and whatnot.

 

I wasn't taking offense at the methodology, just pointing out and quantifying this relevant caveat so the discussion can be a little more precise.

 


Yup, as you say equalizing headphones present some particular challenges as compared to speakers, but I suspect the impact variances of our outer ear, ear canal, cochlea and brain have on our perception of sound mostly overwhelms that of the rest of our body. If it didn't, how could headphones and IEMs work as well for sound reproduction as they do?

post #612 of 1002
It really isn't complicated... it's very simple.

Assume that all my life with the shape of my particular noggin, I've experienced a 10dB boost at 5kHz... and you with yours have experienced a 10dB cut at 5kHz.

If we were both given a recording and were told to "make it sound natural", we would end up in *exactly* the same place, because I would EQ it flat so it would sound like a 10dB boost and you would EQ it flat so it sounded like a 10dB cut to you. "Natural" to us is what we've heard all our lives. It doesn't matter that your "natural" is different than my "natural". We both are hearing the same thing.

That is the nice thing about EQing by ear. You don't need to worry about compensation curves and the shape of your ear canal. You just make it sound *right*. EQing by ear isn't as accurate perhaps, because of the broader tolerance for error than using microphones. But since you're going to be listening to music with ears with the exact same broader tolerance, you don't need to worry about the error.

I find that equalization is the BEST THING you can do to improve the quality of sound in any system. But very few people take advantage of the benefits EQ can provide. Instead, they worry about minute (read: inaudible) differences between cables and amps and ignore equalization completely.

Why is this? It's simple... The reason is because tech heads make equalization sound much more complicated than it's worth. All of the arguments about compensation curves, dummy heads, ear canals and measurement microphones scare people off. They end up accepting imbalances or constantly upgrading randomly in search of a better response curve out of the box.

If folks knew that equalization isn't difficult and it didn't require a lot of testing equipment, perhaps they would give it a try. Learning to EQ by ear is primarily training your ear to recognize incremental improvements in sound quality by using natural sounding recordings as a baseline. I would think that hifi nuts would find that to be fun. Isn't that what this hobby is all about?
post #613 of 1002
Perhaps it's different with in ear monitors because how the little bit of plastic fits in my ear hole is different than the way it fits in yours. I don't own in ear monitors myself, but I am guessing that doesn't matter because sharing another person's IEM would be like sharing their toothbrush. Who wants to mix up someone else's earwax with their own?!
post #614 of 1002
Quote:
Originally Posted by bigshot View Post

It really isn't complicated... it's very simple.

Assume that all my life with the shape of my particular noggin, I've experienced a 10dB boost at 5kHz... and you with yours have experienced a 10dB cut at 5kHz.

If we were both given a recording and were told to "make it sound natural", we would end up in *exactly* the same place, because I would EQ it flat so it would sound like a 10dB boost and you would EQ it flat so it sounded like a 10dB cut to you. "Natural" to us is what we've heard all our lives. It doesn't matter that your "natural" is different than my "natural". We both are hearing the same thing.

That is the nice thing about EQing by ear. You don't need to worry about compensation curves and the shape of your ear canal. You just make it sound *right*. EQing by ear isn't as accurate perhaps, because of the broader tolerance for error than using microphones. But since you're going to be listening to music with ears with the exact same broader tolerance, you don't need to worry about the error.

I find that equalization is the BEST THING you can do to improve the quality of sound in any system. But very few people take advantage of the benefits EQ can provide. Instead, they worry about minute (read: inaudible) differences between cables and amps and ignore equalization completely.

Why is this? It's simple... The reason is because tech heads make equalization sound much more complicated than it's worth. All of the arguments about compensation curves, dummy heads, ear canals and measurement microphones scare people off. They end up accepting imbalances or constantly upgrading randomly in search of a better response curve out of the box.

If folks knew that equalization isn't difficult and it didn't require a lot of testing equipment, perhaps they would give it a try. Learning to EQ by ear is primarily training your ear to recognize incremental improvements in sound quality by using natural sounding recordings as a baseline. I would think that hifi nuts would find that to be fun. Isn't that what this hobby is all about?


I think all of us here are on the same page about this, but there is an interesting distinction between headphones and speakers regarding this method, at least theoretically.

I'm sure you are aware of the corporal effect of low frequencies, how they're picked up by your body and travels through flesh and bone to your auditory system.
This is mostly conjecture, but it seems reasonable to assume that different body types do this in slightly different ways. Big vs. small, short vs tall, fat vs. muscle etc.
I'm pretty scrawny, and suspect my body therefore to be less effective in picking up these frequencies. This doesn't matter anything when I'm equalizing speakers. As long as I concentrate on making them sound the same as live sounds they will sound like live sounds to anyone, as you've pointed out. But what happens when I'm equalizing a set of headphones?
With headphones you would be bypassing that corporal low frequency gain, and therefore feel the need to compensate somewhat with a low end lift in EQ. As I with my skinny body am accustomed to less low end, and therefore inclined to compensate less, my idea of flat would seem somewhat bass-light to someone with a bigger body. That is if my hypothesis holds.
 

90% conjecture this, and I suspect the effect of this would be small to negligible, but you must admit that there at least possibilities for something here?

And mikeaj, please let me know if I interpreted your post correctly.

post #615 of 1002
Quote:
Originally Posted by limpidglitch View Post
 

And mikeaj, please let me know if I interpreted your post correctly.

 

I'm not sure if I'm reading you correctly either, but if the focus is just on missing bass response through the body, that's a separate issue (not what I'm talking about). I'm talking about response at the eardrum. You can read about the so-called "missing 6 dB effect" from headphones not giving you bass response through the body in various papers and discussions. Some people think it's valid to compensate on headphones by boosting bass; others don't and say the lack of bass felt through the body is how it is with headphones and can't / shouldn't be addressed. I'm sure there are other thoughts on it too.

 

 

Quote:
Originally Posted by bigshot View Post

If we were both given a recording and were told to "make it sound natural", we would end up in *exactly* the same place, because I would EQ it flat so it would sound like a 10dB boost and you would EQ it flat so it sounded like a 10dB cut to you. "Natural" to us is what we've heard all our lives. It doesn't matter that your "natural" is different than my "natural". We both are hearing the same thing.
 

No, this is not true. 

 

Let me back up. Why are all relatively neutral headphones actually balanced roughly like this (raw response)

with the huge treble spike 2-4 kHz or so? That's because strapping speakers to your head essentially bypasses the gain of your ears that they naturally have (based on the shape) when sounds are coming from a distance. Hence, you need to compensate for that missing anatomical gain  by boosting the headphone response accordingly.

 

A reasonable explanation is here, under "What is diffuse-field equalization?" at least in the first few paragraphs:

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

 

Though many argue about the correct target and think that DF is a little bright on most recordings and so on.

 

 

A flat response out in free space or in a room (as can be measured by a microphone sitting out there too) gets colored by our anatomy. The colored version is what we hear as neutral, because it's what we're used to. That's the reference point. This reference point is thus different for each person. Let's say the response is R1(f) for you and R2(f) for me. Define it as the deviation from flat in dB at every audio frequency f. R1(f) in your eardrum sounds flat to you, and R2(f) in my eardrum sounds flat to me. So 0 would be flat. If you take speakers and equalize it so it sounds like R1(f) to you, what you've done is moved the response in the air to 0. Hence, when I listen, it will sound like R2(f) and sound flat to me.

 

Headphone response is also colored by our anatomy, just to a lesser extent—importantly, in a different way than the sounds from farther away are. Call the response, the interaction from flat from a certain headphone, as H1(f) for you and H2(f) for me. R1(f) is not equal to H1(f), and R2(f) is not equal to H2(f). If you equalize headphones to sound like R1(f) at your eardrum, it will essentially be producing a sound represented by R1(f) - H1(f). In other words, when you listen to it, you get [R1(f) - H1(f)] + H1(f), which produces R1(f), which sounds right to you. When I listen to it, I get [R1(f) - H1(f)] + H2(f), which is likely different than R2(f). That's the issue. If R1(f) - H1(f) + H2(f) - R2(f) were say constant across different frequencies, no problem. However, it's going to be a little off and a different value at different frequencies. e.g. if it's 2 dB at 2 kHz and -3 dB at 2.7 kHz, that would not be good.

 

This kind of glosses over some complexities and details, but hopefully not important ones. Also, I hopefully didn't make any mistakes.

 

 

Quote:
Originally Posted by bigshot View Post

I find that equalization is the BEST THING you can do to improve the quality of sound in any system. But very few people take advantage of the benefits EQ can provide. Instead, they worry about minute (read: inaudible) differences between cables and amps and ignore equalization completely.
 

Agreed. EQ helps and I do it too and by ear. I'm not espousing a correct method, but I personally check response graphs if available for a general idea and then run sweeps to check the rest, especially in the treble. Getting it a little wrong isn't the end of the world either.

 

Quote:
Originally Posted by bigshot View Post

Perhaps it's different with in ear monitors because how the little bit of plastic fits in my ear hole is different than the way it fits in yours. I don't own in ear monitors myself, but I am guessing that doesn't matter because sharing another person's IEM would be like sharing their toothbrush. Who wants to mix up someone else's earwax with their own?!
 

The sleeves that seal into one's ear canals can easily be removed and replaced, so that's technically not a huge issue maybe. Say if you were giving somebody your set to try, you would put on different tips first. But yes, the geometry and materials of the sleeves makes a significant difference on the sound, as does insertion depth and related issues. It's just like how different positioning of headphones on the head or speakers in a room (or listener position relative to speakers, what with off-axis response being not the same and more) makes a difference.


Edited by mikeaj - 4/7/14 at 2:17pm
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Sound Science
Head-Fi.org › Forums › Equipment Forums › Sound Science › Skeptico Saloon: An Objectivist Joint