Head-Fi.org › Forums › Equipment Forums › Sound Science › Frequency Response Curves
New Posts  All Forums:Forum Nav:

Frequency Response Curves

post #1 of 28
Thread Starter 

So I'm a bit puzzled with frequency response curves.

 

http://www.headphone.com/learning-center/about-headphone-measurements.php

 

If the ideal response curve for an audio device is a flat line for the lows and mids and a downward slope for the highs, why is it that headphones with oddly shaped response curves sound good?

 

 

Based on this graph, one would expect the M50's or the HFI-580's to sound ideal, yet the 580's have typically been regarded as more colored headphones compared to the M50's. Also, though the Ultrasone PRO 550 has a really weird curve, I've read a bunch of reviews around the forums praising their sound quality and clarity.

post #2 of 28

There are two main issues here:

 

First, it's very difficult to accurately measure the frequency response of headphones. You can use a "HATS" (Head And Torso Simulator) but every human has different shaped ears, and ear canals. And when the entire "environment" a headphone is operating in (especially an IEM or earbud) becomes just the ear, it's a lot like putting different speakers in very different rooms--it will significantly change their sound. So if you use the same HATS for all your measurements, and if you always position the headphones exactly the same way, and if you keep everything else the same, you can reasonably compare the measurements to each other. You still can't compare them to what your particular ears and brain will hear, or even to "flat".

 

Second, subjective preferences run wide. Some like lots of bass, some like bright exaggerated high frequencies and many like both--that's why lots of headphones have what many call a "V" shaped response curve--it rises at both ends. It makes the sound more exciting for many people. Others like more accurate sound--look at the response of Etymotic headphones as one example. Someone who buys say Monster Turbines is likely to hate Etymotics and visa versa. It's all about subject preferences.

post #3 of 28

Here is where this article comes in: Headwize - Judging Headphones For Accuracy

post #4 of 28
Thread Starter 
Quote:
Originally Posted by nwavguy View Post

There are two main issues here:

 

First, it's very difficult to accurately measure the frequency response of headphones. You can use a "HATS" (Head And Torso Simulator) but every human has different shaped ears, and ear canals. And when the entire "environment" a headphone is operating in (especially an IEM or earbud) becomes just the ear, it's a lot like putting different speakers in very different rooms--it will significantly change their sound. So if you use the same HATS for all your measurements, and if you always position the headphones exactly the same way, and if you keep everything else the same, you can reasonably compare the measurements to each other. You still can't compare them to what your particular ears and brain will hear, or even to "flat".

 

Second, subjective preferences run wide. Some like lots of bass, some like bright exaggerated high frequencies and many like both--that's why lots of headphones have what many call a "V" shaped response curve--it rises at both ends. It makes the sound more exciting for many people. Others like more accurate sound--look at the response of Etymotic headphones as one example. Someone who buys say Monster Turbines is likely to hate Etymotics and visa versa. It's all about subject preferences.

Ah I see, that seems to make sense. I guess finding a 'good sounding' headphone really is a subjective argument since I have a pair of Sennheiser CX-300 IEM and I find them to be a bit too sibilant while other people find them to be super awesome. I decided to join this community mainly because of that problem actually. When I purchased the CX-300's I mainly read the amazon.com reviews and regarded the 1000+ 4.5/5 star customer reviews as positive ones so I purchased them.
 

 



Quote:
Originally Posted by ClieOS View Post

Here is where this article comes in: Headwize - Judging Headphones For Accuracy

Wow, that was an interesting reading, especially the bass section. I glanced through someone's review of a pair of headphones describing the soundstage effect (I forgot his name, but he's been mentioned in the forums) and found that to be interesting too.

 

How do audio engineers produce the 3-D soundstage in headphones anyhow? It seems like it would be a rather difficult task considering the headphones are only left and right speakers.
 

 


Edited by miceblue - 3/9/11 at 6:05pm
post #5 of 28
Quote:
Originally Posted by miceblue View Post


Ah I see, that seems to make sense. I guess finding a 'good sounding' headphone really is a subjective argument since I have a pair of Sennheiser CX-300 IEM and I find them to be a bit too sibilant while other people find them to be super awesome. I decided to join this community mainly because of that problem actually. When I purchased the CX-300's I mainly read the amazon.com reviews and regarded the 1000+ 4.5/5 star customer reviews as positive ones so I purchased them.

I'm gonna go with "no" here. In so much as headphones are sound *reproduction* and therefore the "best sounding" will be "closest to original".

 

That said: while "flat response" is a laudable goal in the speaker world: it's a bit more complex in the headphone world because of the differences in how music is recorded and how headphones play it back. In specific: music is generally recorded without someone's head (or a faximilie thereof) present. When you are in a real room, your head acts as a selective baffle, as does the shape of your ear.

 

When replayed on a speaker: your head is still there and still performs the same function it would "live". Therefore a speaker reproducing what was in the room at the recording is good.

 

When headphones are used: your head is no longer a baffle (each ear is individually sourced), and your ear shape doesn't have the same effect (because the sound's origin is different both from "original" and from ear-to-ear). So what would be ideal for speakers is a problem for headphones.
 

 

post #6 of 28
Thread Starter 
Quote:
Originally Posted by JerryLove View Post

I'm gonna go with "no" here. In so much as headphones are sound *reproduction* and therefore the "best sounding" will be "closest to original".

 

That said: while "flat response" is a laudable goal in the speaker world: it's a bit more complex in the headphone world because of the differences in how music is recorded and how headphones play it back. In specific: music is generally recorded without someone's head (or a faximilie thereof) present. When you are in a real room, your head acts as a selective baffle, as does the shape of your ear.

 

When replayed on a speaker: your head is still there and still performs the same function it would "live". Therefore a speaker reproducing what was in the room at the recording is good.

 

When headphones are used: your head is no longer a baffle (each ear is individually sourced), and your ear shape doesn't have the same effect (because the sound's origin is different both from "original" and from ear-to-ear). So what would be ideal for speakers is a problem for headphones.
 

 

Hm that's an interesting explanation; I never thought of it that way. So the ideal sound for headphones can depend on the user's ear shape?
 

 

post #7 of 28
Quote:
Originally Posted by JerryLove View Post

I'm gonna go with "no" here. In so much as headphones are sound *reproduction* and therefore the "best sounding" will be "closest to original".

 

That said: while "flat response" is a laudable goal in the speaker world: it's a bit more complex in the headphone world because of the differences in how music is recorded and how headphones play it back. In specific: music is generally recorded without someone's head (or a faximilie thereof) present. When you are in a real room, your head acts as a selective baffle, as does the shape of your ear.

 

When replayed on a speaker: your head is still there and still performs the same function it would "live". Therefore a speaker reproducing what was in the room at the recording is good.

 

When headphones are used: your head is no longer a baffle (each ear is individually sourced), and your ear shape doesn't have the same effect (because the sound's origin is different both from "original" and from ear-to-ear). So what would be ideal for speakers is a problem for headphones.
 

 



Which is why we depend on different headphone equalizations, diffuse field seeming to be the most popular (it's what headroom uses).

post #8 of 28

how is a FR test done, with what sound that is...white noise or a pure tone sweep from 1Hz-20KHz  am thinking that the later is almost worthless as music is lots of freqs all at once aye?  wouldn't really tell anything from a single freq with regards to how 'good' a driver is i guess.

 

apologies if it's not the latter.

post #9 of 28

Quote:
Originally Posted by googleborg View Post

how is a FR test done, with what sound that is...white noise or a pure tone sweep from 1Hz-20KHz  am thinking that the later is almost worthless as music is lots of freqs all at once aye?  wouldn't really tell anything from a single freq with regards to how 'good' a driver is i guess.

 

apologies if it's not the latter.



Depends, but a tone sweep works just fine.  The idea is to measure the amplitude of a frequency in comparison to other frequencies.  If you're worried about harmonics that's THD, if you're worried about transients you're looking more towards square wave tests.  You can't just say because something has a great FR that all other aspects will immediately follow (though in headphones, IME, that's usually the case).

post #10 of 28

ah cheers :)

post #11 of 28
Quote:
Originally Posted by googleborg View Post

how is a FR test done, with what sound that is...white noise or a pure tone sweep from 1Hz-20KHz  am thinking that the later is almost worthless as music is lots of freqs all at once aye?  wouldn't really tell anything from a single freq with regards to how 'good' a driver is i guess.

 

apologies if it's not the latter.


You would need a series simultaneous of tones to measure IMD, though I don't think I've ever seen someone measure that on a headphone.

post #12 of 28
Quote:
Originally Posted by maverickronin View Post


You would need a series simultaneous of tones to measure IMD, though I don't think I've ever seen someone measure that on a headphone.


Actually ryu has done hundreds of such measurements, e.g. PX100:

 

px100_ds.gif

 

From a quick look you can tell that the grados are the same or worse. Most other headphones look fine imo, so nothing to really worry about.

 


Edited by xnor - 3/13/11 at 5:04am
post #13 of 28

Oh, that really is what they are.

 

I just skipped past a lot of the parts I didn't recognize on sight or wasn't sure about because my Japanese sucks and I didn't want to remember anything that might be wrong or misinterpreted.


Edited by maverickronin - 3/13/11 at 5:14am
post #14 of 28
Quote:
Originally Posted by googleborg View Post

how is a FR test done, with what sound that is...white noise or a pure tone sweep from 1Hz-20KHz  am thinking that the later is almost worthless as music is lots of freqs all at once aye?  wouldn't really tell anything from a single freq with regards to how 'good' a driver is i guess.

 

apologies if it's not the latter.



A lot of them use an impulse, which is every frequency at once.

post #15 of 28
Quote:
Originally Posted by MrGreen View Post

A lot of them use an impulse, which is every frequency at once.


Nah not really, an impulse is not a good choice as excitation signal for speakers for various reasons like low frequency inaccuracy or low SNR.

Like Shike said, swept-sine is the way to go. In fact it's an optimal excitation signal for headphones. The actual impulse response is derived mathematically.

 


Edited by xnor - 3/13/11 at 8:45am
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Sound Science
Head-Fi.org › Forums › Equipment Forums › Sound Science › Frequency Response Curves