Chinese / Asian Brand IEM Info Thread
Dec 27, 2017 at 9:28 AM Post #14,011 of 33,556

SomeGuyDude

Headphoneus Supremus
Joined
Jun 24, 2012
Posts
7,308
Likes
2,932
As far as I can see that graph doesn't even have any scale on the axis, kind of says everything about it imo......

The proportions are more important than the axis scale. If I showed you an FR graph for any major headphone even without scale you'd get a pretty good idea of what you're looking at.
 
Dec 27, 2017 at 10:12 AM Post #14,012 of 33,556

FUYU

500+ Head-Fier
Joined
Oct 20, 2014
Posts
947
Likes
893
Location
DE
The proportions are more important than the axis scale. If I showed you an FR graph for any major headphone even without scale you'd get a pretty good idea of what you're looking at.

Not really. You'd assume that the respective axes (X and Y) are linear, which often is not the case, particularly with Chinese manufacturers. Frequency response is not really a good indicator without a reference. Even when measuring the same earphones, there is lots of variance to account for.
 
Dec 27, 2017 at 10:16 AM Post #14,013 of 33,556

SomeGuyDude

Headphoneus Supremus
Joined
Jun 24, 2012
Posts
7,308
Likes
2,932
Not really. You'd assume that the respective axes (X and Y) are linear, which often is not the case, particularly with Chinese manufacturers. Frequency response is not really a good indicator without a reference. Even when measuring the same earphones, there is lots of variance to account for.

Yes really, because no matter the scale the relationship between two points along the X axis is going to be the same. If the output at 100Hz is the same as the output at 5000Hz it doesn't matter what the scale is, they're going to be at the same Y coordinate. That's literally how graphs work.
 
Dec 27, 2017 at 10:36 AM Post #14,014 of 33,556

FUYU

500+ Head-Fier
Joined
Oct 20, 2014
Posts
947
Likes
893
Location
DE
Yes really, because no matter the scale the relationship between two points along the X axis is going to be the same. If the output at 100Hz is the same as the output at 5000Hz it doesn't matter what the scale is, they're going to be at the same Y coordinate. That's literally how graphs work.

Yes, that's how a function is projected on a Cartesian coordinate system. "If I showed you an FR graph for any major headphone even without scale", but if I were to change that scale or the ratio of the Y-Axis the entire projection would flatten or elevate. It changes your outlook of the graph entirely, especially when you have no scaling to begin with. And I don't think you look at the points of the graph individually.
 
Dec 27, 2017 at 11:23 AM Post #14,015 of 33,556

SomeGuyDude

Headphoneus Supremus
Joined
Jun 24, 2012
Posts
7,308
Likes
2,932
Yes, that's how a function is projected on a Cartesian coordinate system. "If I showed you an FR graph for any major headphone even without scale", but if I were to change that scale or the ratio of the Y-Axis the entire projection would flatten or elevate. It changes your outlook of the graph entirely, especially when you have no scaling to begin with. And I don't think you look at the points of the graph individually.

It changes amplitude, not proportions, and proportions are the primary thing we're looking at, especially when it comes to deciding if it's warm/bright/neutral.
 
Dec 27, 2017 at 11:34 AM Post #14,016 of 33,556

FUYU

500+ Head-Fier
Joined
Oct 20, 2014
Posts
947
Likes
893
Location
DE
It changes amplitude, not proportions, and proportions are the primary thing we're looking at, especially when it comes to deciding if it's warm/bright/neutral.
Yes, that is exactly my point. The projection and its amplitude changes. Hence, it manipulates the graph to possibly look more/less U, V, etc.-shaped. Wasn't this the whole point of it? To make an accurate assumption of the frequency response and subsequent tonality?
 
Last edited:
Dec 27, 2017 at 11:37 AM Post #14,017 of 33,556

SomeGuyDude

Headphoneus Supremus
Joined
Jun 24, 2012
Posts
7,308
Likes
2,932
Yes, that is exactly my point. The projection and its amplitude changes. Hence, it manipulates the graph to possibly look more U, V, etc.-shaped. Wasn't this the whole point of it? To make a accurate assumption of the frequency response and subsequent tonality?

No, because all this started thanks to the King Pro's frequency graph which looked like this:

HTB1j89mj8DH8KJjSszcq6zDTFXaa.jpg


Your talk about U or V shapes is just a matter of the extent of what is a symmetrical relationship. If a headphone starts with bass way high and then falls off consistently from there it's a bassy headphone no matter what the scale is. Likewise, this here graph shows a damn worrisome bump regardless of what the exact axes are.
 
Dec 27, 2017 at 11:53 AM Post #14,018 of 33,556
Joined
Sep 12, 2012
Posts
10,899
Likes
14,551
Location
Norway
No, because all this started thanks to the King Pro's frequency graph which looked like this:

HTB1j89mj8DH8KJjSszcq6zDTFXaa.jpg


Your talk about U or V shapes is just a matter of the extent of what is a symmetrical relationship. If a headphone starts with bass way high and then falls off consistently from there it's a bassy headphone no matter what the scale is. Likewise, this here graph shows a damn worrisome bump regardless of what the exact axes are.

If the interval on the y-axis is 1dB the bumb is pretty neglecable, if the interval on the x-axis is big it's even more so.

My point is still that as long as the intervals are unknown the graph is useless. I realize that you'd like to argue otherwise but I honestly cannot see how, I've only studied statistics for a couple of years and that was many moons ago though so maybe I'm missing something....
 
Dec 27, 2017 at 11:57 AM Post #14,019 of 33,556

SomeGuyDude

Headphoneus Supremus
Joined
Jun 24, 2012
Posts
7,308
Likes
2,932
If the interval on the y-axis is 1dB the bumb is pretty neglecable, if the interval on the x-axis is big it's even more so.

My point is still that as long as the intervals are unknown the graph is useless. I realize that you'd like to argue otherwise but I honestly cannot see how, I've only studied statistics for a couple of years and that was many moons ago though so maybe I'm missing something....

You're the one being pointlessly argumentative here because it's flamingly obvious what's going on in the chart but you're being all "well what if the scale is different than every other graph? What if the guy making it was drunk? What if they dropped the chart in brown sauce and had to redraw it from memory?"

You can see what's happening there, quit playing dumb games.
 
Dec 27, 2017 at 12:00 PM Post #14,020 of 33,556
Joined
Sep 12, 2012
Posts
10,899
Likes
14,551
Location
Norway
You're the one being pointlessly argumentative here because it's flamingly obvious what's going on in the chart but you're being all "well what if the scale is different than every other graph? What if the guy making it was drunk? What if they dropped the chart in brown sauce and had to redraw it from memory?"

You can see what's happening there, quit playing dumb games.

Eh, maybe you should go out for walk or something......

I'm not the one spending post after post on this meaningless argument.
 
Last edited:
Dec 27, 2017 at 12:09 PM Post #14,021 of 33,556

FUYU

500+ Head-Fier
Joined
Oct 20, 2014
Posts
947
Likes
893
Location
DE
No, because all this started thanks to the King Pro's frequency graph which looked like this:

Pardon, "To make a accurate assumption of the frequency response and subsequent tonality of the King Pro". Yes, if the Axis is "linear" (I forgot the terminology here) like going from 1 to 2 to 3, or 3 to 6 to 9, or log1 to log0 to log-1, etc.. But there have been instances where the scaling suddenly jumps from 100 to 110 to 130 for no reason, just to alter the projection, making it appear more balanced/whatever.

Frequency response is not really a good indicator without a reference. Even when measuring the same earphones, there is lots of variance to account for.

My actual point was that a singular graph describing tonality is nigh useless, without having A) Metrics for what you are actually measuring B) What measuring tool/standards you apply C) Model the graph with compensation and/or factors like insertion depth, tip-selection, etc. and D) having a reference plane.
Even though A) is SPL(dB)(Y-Axis) and Frequency in hz (X-Axis) in 99% of cases, it still could be something entirely different for what its worth.
 
Last edited:
Dec 27, 2017 at 1:20 PM Post #14,023 of 33,556

HungryPanda

Headphoneus Supremus
Joined
Dec 8, 2013
Posts
7,974
Likes
13,282
Location
Kent, UK
Yes George, only ones ears can actually tell what they sound like. People could argue all day about nothing I suppose :wink:
 

Users who are viewing this thread

Top