headphone tonality test

[Thomas_A]

The variations would be that that person does not have a full grasp of the reference tonality of their headphone or IEM. If you know how your IEM / headphone sounds from 20Hz - 20KHz, you'll easily within 5 seconds that #5 is the original and everything is a deviation from the reference tonality of your chosen IEM / headphone (even if you EQed headphones to Harman and your EQed headphones / IEMs to Harman will be your reference point and any deviations from it will be easily noticed). I posted my results before @solderdude showed the results. Clearly most responses knew how their system sounds and as such they can easily tell the original from manipulated

If you have a reference yes but without such reference you rely on your inner reference of how eg a voice would sound. It requires that the original has a balanced sound. Relative deviations are much easier to detect.

 

[majingotan]

If you have a reference yes but without such reference you rely on your inner reference of how eg a voice would sound. It requires that the original has a balanced sound. Relative deviations are much easier to detect.

That's another way to get the same results. It'll be fun for a casual listener (those who aren't in to audio hobby) use a headphone they have zero idea of how it sounds like (because they won't have a reference point) and now rank the samples. Closest to a double blind test and thus would have no bias and results would be very interesting

 

[Thomas_A]

That's another way to get the same results. It'll be fun for a casual listener (those who aren't in to audio hobby) use a headphone they have zero idea of how it sounds like (because they won't have a reference point) and now rank the samples. Closest to a double blind test and thus would have no bias and results would be very interesting

If you e.g. would use Springsteens album ”Born in the USA” to anyone that have not heard the album before, you might end up with more bass boost judged as the original.

 

[majingotan]

If you e.g. would use Springsteens album ”Born in the USA” to anyone that have not heard the album before, you might end up with more bass boost judged as the original.

If I had listen to it on a headphone that I've never auditioned before such as the Sprit Torino Valkyria then probably I might've chosen that instead of the original since I don't know its exact tonality and there's no published FR curve for that headphone AFAIK so no chance of having a reference point

 

[MayaTlab]

If you have a reference yes but without such reference you rely on your inner reference of how eg a voice would sound. It requires that the original has a balanced sound. Relative deviations are much easier to detect.

Agreed, there's a bit of a leap of faith here. But for some of the tracks (2, 3) there were to me some specific medium-ish Q issues that I would not expect from what seemed like a reasonably well recorded / mastered track otherwise (I mean it didn't sound like this was incompetently captured or mastered, or intentionally coloured I guess ?), and that got the ball rolling to end up after a while thinking that 5 seemed to me the most likely candidate to be the original track. For two tracks (3,4) I also was already very familiar with how the sort of deviations these headphones produce vs. a more "normal" curve sound like with "normal" recordings. So assuming that 5 was the original, I then proceeded by elimination until the whole picture made some sense to me. Had a more coloured track been used I am not certain that I would have been able to form an opinion on which track was the original.

 

[Thomas_A]

If I had listen to it on a headphone that I've never auditioned before such as the Sprit Torino Valkyria then probably I might've chosen that instead of the original since I don't know its exact tonality and there's no published FR curve for that headphone AFAIK so no chance of having a reference point

I am with you on that. What I am trying to say is that even if you have a familiar headphone or tuned to Harman you might still get wrong on what is an original if the track you are listening to is unknown to you. You have to rely on your inner reference and that the mix hopefully was balanced.

 

[MayaTlab]

You have to rely on your inner reference and that the mix hopefully was balanced.

I agree. I am not certain that listening tests would amount to much with this sort of track .

 

[Chagall]

Agreed, there's a bit of a leap of faith here. But for some of the tracks (2, 3) there were to me some specific medium-ish Q issues that I would not expect from what seemed like a reasonably well recorded / mastered track otherwise (I mean it didn't sound like this was incompetently captured or mastered, or intentionally coloured I guess ?), and that got the ball rolling to end up after a while thinking that 5 seemed to me the most likely candidate to be the original track. For two tracks (3,4) I also was already very familiar with how the sort of deviations these headphones produce vs. a more "normal" curve sound like with "normal" recordings. So assuming that 5 was the original, I then proceeded by elimination until the whole picture made some sense to me. Had a more coloured track been used I am not certain that I would have been able to form an opinion on which track was the original.

Nice! You got everything right.

Now thinking about it there was a simple way to know the original. We knew that tonalities were from headphones reviewed here and most of them need a bass boost (except DCAs), so any that lack bass can't be the original 2, 4, and certainly 3. That leaves 1 or 5 with 1 being boomy with no treble - 5 was the original.

Fun test though.

 

[OnLyTNT]

If you have a reference yes but without such reference you rely on your inner reference of how eg a voice would sound. It requires that the original has a balanced sound. Relative deviations are much easier to detect.

That's the point I failed miserably. I took the most balanced one as my reference according to Hexa. Also, I shouldn't use Hexa for the task at the first place I guess .

 

[muza_1]

That's the point I failed miserably. I took the most balanced one as my reference according to Hexa. Also, I shouldn't use Hexa for the task at the first place I guess .

At first I tried with my EQ on the Zero:2 (Harman 2019v2 plus a LF/105Hz/+1.5dB and 6kHz/-2dB/Q2.0) and it was difficult to identify the differences and tonal balance of the original (mostly in the low-end region), in the end I just used the Harman target. I assumed it was a good recording form the start if not it would have been almost impossible to know which one was the original.

 

[solderdude]

The 'trick' here was to assume the Harman target was a valid target.
The recording was (I assume) mixed on speakers in a studio so it might have been best to listen to the files on well extended speakers.
As headphones need a bit of a bass boost to 'mimic' the impact of listening to speakers in a decent listening room which the Harman OE curve is supposed to mimic, I simply EQ'ed the deviations into the files.

The recording itself was a bit bassy, probably to create a bit of the 'live feel' during the recording ?
I was unsure to name the reference 'reference' which would perhaps make the comparison fairer but not doing so would show if people would prefer the original (5) over a different tonal balance.

 

[muza_1]

The 'trick' here was to assume the Harman target was a valid target.
The recording was (I assume) mixed on speakers in a studio so it might have been best to listen to the files on well extended speakers.
As headphones need a bit of a bass boost to 'mimic' the impact of listening to speakers in a decent listening room which the Harman OE curve is supposed to mimic, I simply EQ'ed the deviations into the files.

The recording itself was a bit bassy, probably to create a bit of the 'live feel' during the recording ?
I was unsure to name the reference 'reference' which would perhaps make the comparison fairer but not doing so would show if people would prefer the original (5) over a different tonal balance.

I did not say in my original post but I do prefer the original over the others.

 

[OnLyTNT]

What is the original track by the way?

 

[solderdude]

That makes sense as you EQ'ed to Harman anyway leveling the playing field.
It shows that Harman tonality is what you prefer (and use as a reference)
This way you kind of sampled the Stealth, Caldera, Susvara and HD650 'as is' according to Amir's measurements.

 

[solderdude]

What is the original track by the way?

There are many more AVO sessions that are worth listening/seeing.
It shows how good artists really are when playing live.

 

[muza_1]

That makes sense as you EQ'ed to Harman anyway leveling the playing field.
It shows that Harman tonality is what you prefer (and use as a reference)
This way you kind of sampled the Stealth, Caldera, Susvara and HD650 'as is' according to Amir's measurements.

I find the Harman target a bit bright and fatiguing even if it is the correct tonality and I like a bit more of bass about a 1dB and because I mostly use them outside I add another 1dB more to compensate for ambient noise.

 

[solderdude]

Harman is a bit too bright for me too (too much ear-gain) and I like 2 dB less bass boost and a different slope from 20Hz to 200Hz (closer to DCA but less elevated).
This is when comparing to reality and properly EQ'ed near fields.
May be an outer-ear thing differing from a KB5000 pinna ?

 

[OnLyTNT]

There are many more AVO sessions that are worth listening/seeing.
It shows how good artists really are when playing live.

Ahh, I noticed it was Mark Knopfler but not a hardcore fan of him, I couldn't pin point the song. Thanks.

 

[muza_1]

Harman is a bit too bright for me too (too much ear-gain) and I like 2 dB less bass boost and a different slope from 20Hz to 200Hz (closer to DCA but less elevated).
This is when comparing to reality and properly EQ'ed near fields.
May be an outer-ear thing differing from a KB5000 pinna ?

It is incredible how close we get to a natural/balanced response with just a few minor differencces considering that measurment equipement uses an averaged pinna and ear canal and the difficulties to measure.

Maybe they should compensate for treble in the same way they do for bass (not by measurement but to mimic the impact as you said) for headphone and especially IEMs targets, way too many people find the IEM Harman target bright.

 

[solderdude]

The 'problem' with a single and overly 'smoothed' target that is declared holy on 1 specific fixture is that in the Harman research it became obvious that there is a substantial 'bandwidth' in preference. If Harman would publish tolerance bands along side 'the target' and used color coded bands (say red is what most people got and blue what outliers got) then it would become obvious there was no single Harman curve as there would be a rather wide 'band' below 200Hz and above a few kHz.

That would confuse the hell out of people and Harman would not have something to 'aim' for. So the choice of using a best fit for the average and use a single 'trace' makes a lot of sense as Harman was aiming to get close to a tonality that would appeal to the majority of people. This was the goal in order to make headphones that sold better (based on sound).
Most people, however, buy on looks, price, functions and care less about sound quality and comfort/fit.

Not so for ASR readers who seem to value sound quality (linked to tonality), get all upset about distortion (which isn't as audible as most think it is, and value comfort as usually they listen for longer time periods to enjoy music instead of 'wanting to hear something' to fill in the silence or drown out noises around them in a fashionable way'.