• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Studio headphones guide by sonarworks

pierre

Addicted to Fun and Learning
Forum Donor
Joined
Jul 1, 2017
Messages
962
Likes
3,047
Location
Switzerland
Well written and free, available Here.

Sonarworks is flattening the frequency response at listening position. That's ok for speakers, what about headphones?
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
I think it's O.K. to EQ, and to flatten headphones.

The debate seems to be whether or not measurements are accurate and how they should be made and (when called for) which 'compensation' is needed.
What the 'baseline' should be and who and how determined it.
Whether we want 'speakers in room' emulation or want to reproduce the recording (different tonality)
Also depends on the headphone used and the amount of EQ needed.
Also depends on listening level.

I have my thoughts but there are others that have theirs.
Live and let live.
There is no 'standard' (yet) so everyone is winging it.
As long as it brings sonic improvements I am all for it.
 

andreasmaaan

Master Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
6,652
Likes
9,399
I have my thoughts but there are others that have theirs.

Would be interested to hear your thoughts having listened to and measured so many headphones @solderdude :) Particularly on whether you prefer cans with a room response (Harman-style) or ones that measure a bit flatter?
 

Sancus

Major Contributor
Forum Donor
Joined
Nov 30, 2018
Messages
2,923
Likes
7,616
Location
Canada
They don't seem to mention what HRTF they're using? That seems like important information. How you define 'flat' on a headphone is pretty heavily dependent on what methodology you're using, because the unadjusted frequency response is definitely not targeted to flat on any headphone.
 

flipflop

Addicted to Fun and Learning
Joined
Feb 22, 2018
Messages
927
Likes
1,240
They don't seem to mention what HRTF they're using? That seems like important information. How you define 'flat' on a headphone is pretty heavily dependent on what methodology you're using, because the unadjusted frequency response is definitely not targeted to flat on any headphone.
By "HRTF" I assume you mean 'compensation curve'. They have confirmed elsewhere they're using a modified Harman curve. Comparing Sonarworks measurements to measurements from other sources, it seems their curve have more energy in the bass and around 2 kHz.
They're not very transparent about those things though and despite having made hundreds of measurements, they have no intentions of releasing the data to the public.
 

Sancus

Major Contributor
Forum Donor
Joined
Nov 30, 2018
Messages
2,923
Likes
7,616
Location
Canada
By "HRTF" I assume you mean 'compensation curve'. They have confirmed elsewhere they're using a modified Harman curve. Comparing Sonarworks measurements to measurements from other sources, it seems their curve have more energy in the bass and around 2 kHz.
They're not very transparent about those things though and despite having made hundreds of measurements, they have no intentions of releasing the data to the public.

Yes sorry that's what I meant. Based on my understanding to get a correct compensation curve for you personally you need to have your HRTF measured because it varies from person to person, significantly so.

Taking the Harman curve and adjusting it is essentially just the writers adding their own opinion of what a headphone should sound like, or possibly adding adjustments because their measurement setup is different from Harman's(which I believe also changes things).

If you haven't taken the effort to do that it's not clear to me that typical published headphone measurements are as useful as the ones for speakers in general. But I find the whole conversation about headphone measurements very confusing, and it seems like there's still substantial changes being made from ongoing research.
 

flipflop

Addicted to Fun and Learning
Joined
Feb 22, 2018
Messages
927
Likes
1,240
Yes sorry that's what I meant. Based on my understanding to get a correct compensation curve for you personally you need to have your HRTF measured because it varies from person to person, significantly so.
The measured frequency response between subjects can vary quite a bit in the treble range, however, it's not perfectly clear how much our brains compensate for those differences.
Taking the Harman curve and adjusting it is essentially just the writers adding their own opinion of what a headphone should sound like, or possibly adding adjustments because their measurement setup is different from Harman's(which I believe also changes things).
Yes, that's unfortunately what they seem to be doing.
Rtings.com did something similar, where they corrected the Harman curve to fit their own measurement rig, which was different from the one used by Harman. They also corrected the bass to fit their own preferences, so they actually did both of the things you're describing. At least they made a video explaining the situation:
With Sonarworks, you're left speculating.
If you haven't taken the effort to do that it's not clear to me that typical published headphone measurements are as useful as the ones for speakers in general. But I find the whole conversation about headphone measurements very confusing, and it seems like there's still substantial changes being made from ongoing research.
The Harman target curve predicts listener preference with a 0.91 correlation coefficient, which is higher than the 0.86 loudspeaker measurements provide. I think the main issues with headphone measurements are the differences in testing methodologies and the differences in HTRFs between dummy heads, leading to different results, even when the exact same pair of headphones are being measured.
 

Sancus

Major Contributor
Forum Donor
Joined
Nov 30, 2018
Messages
2,923
Likes
7,616
Location
Canada
The Harman target curve predicts listener preference with a 0.91 correlation coefficient, which is higher than the 0.86 loudspeaker measurements provide. I think the main issues with headphone measurements are the differences in testing methodologies and the differences in HTRFs between dummy heads, leading to different results, even when the exact same pair of headphones are being measured.

Thanks, this is good to know, I didn't realize the correlation was that high for the Harman curve.
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
Would be interested to hear your thoughts having listened to and measured so many headphones @solderdude :) Particularly on whether you prefer cans with a room response (Harman-style) or ones that measure a bit flatter?

My preference (probably because I don't listen very loud) is 'flat' with a mild bass boost.
Not as extreme as Sonarworks nor Harman. Less boost and based on the difference between an 80-85dB mastered recording and 70dB reproduction.
So the bass boost is not as steep as Harman either but more 'natural' 6dB/octave alike.

f-m-overlayed.png


So my compensation used when measuring is below:
corr-plot.png


my explanation is found here.

Personally I don't think HRTF is that much of a problem for headphones. It is for measurement equipment.
The reason I think this is true is that our ears calibrate themselves and take into account where the sounds are coming from.
A violin played in front of you or to the side still sound equally 'violin'.

The compensation needed for dummy heads with ear canals and Pinna is very complex and headphone depenedant.
For this reason (and financial) I prefer to measure what the driver itself puts out, not what 'a' fake Pinna does and 'a' fake ear canal does and later have to undo something complex.
usually my plots end up quite close to 'compensated' dummy head measurements.

I also believe one should not 'correct' to the exact opposite of what is measured as that may differ from someone else's HRTF.
Instead I look for grossser errors that need correction and then apply a gentle EQ to get it in the correct ballpark.

There are a few headphones that measure close to 'flat' and sound good to me without any EQ.

below the first HE-6 for instance.
he6-fr.png


Even here one can see variations well over 5dB (which is audible) if this were the FR of an amp or DAC everyone would be all over it claiming how poor the performance is.
You can actually see my compensation in that plot as bass measures ruler flat but due to my compensation is now slightly on the thin side.

But indeed the vast majority of headphones deviates many dB's, sometimes well over 15dB in specific areas.
This makes them all sound (and measure) very different.
You can get them closer by applying EQ... but one needs to base EQ on something.
And that is indeed where the problem area is...

Others may have quite differing views (such is life) and everyone should use what they prefer or think is right.
I don't claim my measurements are right nor my method is right. I get good and consistent results this way which makes me wander this path.

The Sonarworks 'paper' makes sense (outside the dot of the middle C (261.6Hz) being shown on the C on the left of the text where it should have been on the C on the right of the text.

Also the SHP9500 recommendation is a bit suspect as it does not comply to their own recommendations (replaceable pads).
 
Last edited:

Frank Dernie

Master Contributor
Forum Donor
Joined
Mar 24, 2016
Messages
6,445
Likes
15,780
Location
Oxfordshire
I have AKG Q90s they have a DSP system which compensates for the acoustic space when the actual listener is wearing them. The difference it makes is considerable and a lot better.
 

SIY

Grand Contributor
Technical Expert
Joined
Apr 6, 2018
Messages
10,383
Likes
24,749
Location
Alfred, NY
Sonarworks has a rather different way of doing their headphone measurement (outlined in the Sprogis patents), but at least you can see what their EQ curve is for each model.

Purely subjectively, I was quite impressed with what their EQ did for some cheap headphones I had on hand when I was reviewing their software.
 

JohnYang1997

Master Contributor
Technical Expert
Audio Company
Joined
Dec 28, 2018
Messages
7,175
Likes
18,292
Location
China
I have tried before. The eq is far off. Their speaker correction is quite good but they don't seem to know how headphones work.
 

DDF

Addicted to Fun and Learning
Joined
Dec 31, 2018
Messages
617
Likes
1,355
Purely subjectively, I was quite impressed with what their EQ did for some cheap headphones I had on hand when I was reviewing their software.

I was as well with their eq of my Shure SRH840.
 

Frank Dernie

Master Contributor
Forum Donor
Joined
Mar 24, 2016
Messages
6,445
Likes
15,780
Location
Oxfordshire
On that Sonarworks link, didn't they mislabel their FR curves with 100 khz where 10 khz should be?
https://www.sonarworks.com/static/S...286768713&mc_cid=93ba99d760&mc_eid=f1c63b6fc5
I was surprised by a lot of their frequency based notes. Firstly marking middle C one octave below where it actually is and calling up to 300 Hz upper bass and 800 to 2.5k mid range to me from a musical perspective. If I play thse notes on our piano I wouldn't call middle C in the "upper bass" or 3 octaved higher "mid-range".
Mind you maybe the nomenclature has nothing to do with the sound and is just the way of splitting up human frequency range into bite-sized chunks.
I have an app on my phone SPLnFFT which allows an FFT of the sound to be watched. On the sort of music I listen to the vast bulk of the energy is going on below 1kHz and whilst I wouldn't expect the microphone to have an extended response at the frequency extremes the middle bit is probably fine.
Certainly the full range of overtones makes the sound brighter and need reproducing for "high-fidelity" but, goodness, a fundamental at the top of a grand piano keyboard sounds very high, yet is less than 4kHz.
Certainly the harmonic content of a piano is rich and varying, the fundamental decays steadily but the harmonics vary on a sustained note.
 
Top Bottom