(Sticky?) Use your ears like to state a more objective Frequency Response!
Jun 30, 2008 at 11:53 PM Thread Starter Post #1 of 22

antonyfirst

Headphoneus Supremus
Joined
Dec 14, 2006
Posts
4,051
Likes
54
Location
Milan
This will help lots of us to put our impressions in perspective, and give some more objective information to other people.

Equal loudness contours and audiometry - Test your own hearing

The graph works this way: you play different notches for each frequency, with your headphones on, and you can state the loudness curve of your headphones to your ears.
The lower you go, for each column, the lower volume will be played by your computer. So you, an headphone that lets you hear well at a low value is very sensitive to that volume.
If you flip the resulting curve, you have a frequency response like those shown at Headroom.

Let's do a try:

2626565972_42c21382f5_o.jpg


Like I stated many times before, the RE1 have soft treble (relaxed rather than exciting), suppressed upper midrange (6kHz) which results in muted cymbals (not so good) and total forgiveness to sibilance (good). They have full vocals (a bit chesty sometimes), and fat, deep bass, that wouldn't be out of proportion if they had better upper midrange.

2626566272_12dedbe1d3_o.jpg


I described the OK2 like having great treble, crystal clear vocals and good detail. Some slight sibilance with harsh recordings, but a very flat and enjoyable sound overall. I didn't realize that bass was so severely rolled off. There's no midbass bloat (good), nor shaky deep bass (bad). The latter feature has to be expected for an earbud design.

The graph goes up to 16 kHz, so it's thought better for IEMs/earphones than full size headphones, that usually reach higher.

This is the right way to take a reference volume:
Quote:

Originally Posted by MaloS
In the 1 kHz column, choose a panel about halfway down. Listen to it and check that (i) it is not uncomfortably loud and (ii) it is considerably louder than the background noise. Go up or down the column if necessary until these conditions are satisfied. Notice that the dB level chosen is now recorded at the bottom of the chart. Once you have made a choice, this becomes your reference sound.


Important!!!
To save the image
, press the "Stamp" key on your keyboard, it will take an image of the screen. Then open Paint and paste the screenshot in it. Crop the region of interest and upload it somewhere (photobucket, flickr, please don't use imageshack since many of us are having problems viewing images at their full size).
 
Jul 1, 2008 at 12:00 AM Post #2 of 22
Tony, I don't have time to play with this right now but thanks! Looks a very good resource from my brief scan...I'll have to read it more carefully tomorrow.

Ta very much
smily_headphones1.gif
 
Jul 1, 2008 at 12:21 AM Post #3 of 22
Tony, are you sure you did this test correctly? I have done this and the lowest setting that I heard at my best frequency was about 90 dB below zero. Of course my chart had a similar pattern to the one that you put up having that concave shape.

What I did was to take the lowest level that I could hear at that particular frequency. This way my standard for each frequency was my limit of hearing, otherwise how would I actually know what volume I was listening too?
Hearingtestforstevejan307.png

Jan 07
 
Jul 1, 2008 at 12:26 AM Post #4 of 22
I choose the loudness level arbitrary. I wasn't interested to in what was my maximum hearing, but in the response curve of those earphones. The "absolute volume" is not important. The relative volume is.

(side note: I could hear the 16 kHz at -66 dB with the OK2 with an average volume out of the amp I used)

But it's true that we could use the 0 dB as reference, rather than an arbitrary volume that's lower than that. It should not make a difference, but we might need an expert for further explanations.
 
Jul 1, 2008 at 12:32 AM Post #6 of 22
Quote:

Originally Posted by FlavioWolff /img/forum/go_quote.gif
nice


What's "nice"?
biggrin.gif
biggrin.gif
biggrin.gif

Do it for your modded Sonys, and also with the PK1 when you get them.
biggrin.gif
 
Jul 1, 2008 at 12:37 AM Post #8 of 22
This is the Stax SRS-252a system. I should do the ESW9, but too lazy to go get it at the moment.
ELCStaxSRS-252a.jpg


I followed this when choosing volume for 1khz btw:
Quote:

In the 1 kHz column, choose a panel about halfway down. Listen to it and check that (i) it is not uncomfortably loud and (ii) it is considerably louder than the background noise. Go up or down the column if necessary until these conditions are satisfied. Notice that the dB level chosen is now recorded at the bottom of the chart. Once you have made a choice, this becomes your reference sound.


 
Jul 1, 2008 at 12:43 AM Post #9 of 22
Quote:

Originally Posted by MaloS
In the 1 kHz column, choose a panel about halfway down. Listen to it and check that (i) it is not uncomfortably loud and (ii) it is considerably louder than the background noise. Go up or down the column if necessary until these conditions are satisfied. Notice that the dB level chosen is now recorded at the bottom of the chart. Once you have made a choice, this becomes your reference sound.


OK, then I guessed right and chose the right way to do my curves.
 
Jul 1, 2008 at 8:34 AM Post #12 of 22
I used a pair of Er4P that are only about 5 hrs old and plugged straight into a Powerbook G4.

The first one follows the method of selecting a volume at 1Hz and then comparing relative volumes:
w-er4p.jpg



For fun, I did it a second time with the volume on the laptop set at 50% and then selecting the lowest audible volume I could detect:
lowestaudible.jpg
 
Jul 1, 2008 at 5:56 PM Post #13 of 22
I'm not sure how valuable this really is due to all the sources of possible variances. A difference in level could be due to your hearing, or the capability of the headphone is use, or even the starting level you choose (since human hearing sensitivity at high and low frequencies varies quite a bit with sound level.) Without being able to isolate all these factors I'm not sure what this test proves. At the very least you would need some calibrated reference headphones for it to have any meaning at all as a hearing test, and then you would need a plot of that hearing test to use to evaluate any consumer headphones, and results would be different at different listening levels, etc....
 
Jul 1, 2008 at 6:59 PM Post #14 of 22
You didn't get the point. We aren't testing our hearing, we are testing our headphones.
 
Jul 1, 2008 at 8:40 PM Post #15 of 22
No, you didn't get the point, and testing headphones with this method is even more flawed than trying to test hearing. Human hearing sensitivity varies greatly at high and low frequencies based on level. If you made sure that every listener used the same level settings (and adjusted them for the sensitivity of their headphones) then you might have a start, but even then you're probably just going to get more of the inverted Bell curve that we've been seeing so far. Plus, everyone's hearing is different and a plot for one person listening to a particular headphone may not be all that close to another.

The only meaningful test for headphones would be on a fixture with microphones adjusted for a flat response in that particular environment.
 

Users who are viewing this thread

Back
Top