Here's the problem with this rationale for including your additional "Default Harman curve -1.5dB@ HF":
1. It's a
single arbitrary curve in the 'taste zone' as you put it...
2. at one extreme of the treble taste zone no less, so
unlikely to be applicable to many listeners...
3. based on your 'casual observation' of what
you think is...
4.
one particular listener's preference which applies specifically to them...
5. in their
sighted, non-level matched conditions, so even if correctly inferred by you, may not match with true blinded, level-matched preference...
6. a target which
@Robbo99999 has pointed out
isn't consistently matched in the same way to as your Harman target EQ, as is the case with the EQ for this headphone.
7. All this accompanied by
not scientifically valid adjusted 'scores' with this new target that have no basis in research that specifically uses the Harman target in the algorithm's calculation of the predicted preference score. You can't just change the target in the algorithm and expect that to be valid.
There's just too much arbitrary unscientific inference going on here for this curve (and the sometimes inconsistent EQing to it compared to Harman) to warrant being plucked out of the 'taste zone' and presented alongside the actual Harman target which has strong scientific backing behind it. Casual readers will see it as having some special significance, if not on a par with the Harman target, then at least more than other possible curves in the 'taste zone', which it really doesn't (and the accompanying adjusted scores as being valid predictions, which they're not). An approach such as Oratory's is better here - EQ to the Harman target, but carefully curate the bands to include some broad filters such that regions like bass and treble can be fine-tuned according to preference / headphone unit / listener anatomy / mixing or mastering variation.
1. It's a single arbitrary curve in the 'taste zone' as you put it...
It is not arbitrary, I have explained the rational behind it and if you have had a look at the curves, some versions of the of the target curves had lower output @HF and LF (trained listeners?). Here again:
You can question the details of the rational, I am OK with that, but I don't think you can confidently state that the lower HF target is completely arbitrary.
2. at one extreme of the treble taste zone no less, so unlikely to be applicable to many listeners...
Well, again please read the graphs provided for yourself:
Men tend to forget that half of them are women... Women tend to prefer -3.1dB. At HF alone -1.5dB@HF target (or -1.41 -1.5dB = -2.9dB) might be closer to 50ish% percent of the population preference. So not such a small portion of the population, is it?
In addition the more training the less HF (and LF) is preferred which may or may be representative of the of ASR readership.
The map and the country, or if you prefer, model vs reality.
The users must not comply with the target but the target can be adjusted to the user.
All the models are good
approximations so most probably, in the details, every user has his/her own slightly different preference target.
Again from the study at least a third of the population exhibits significant deviation from the default target.
The whole "taste zone" is the "Harman target curve" for the complete population (with "normal hearing").
It may not matter if we are only ranking devices against each other (although it may, see hereafter) but it will matter when trying to EQ to the exact target. Readers need to understand that and I wrote it in my EQ preamble.
As a matter of fact, to me, one of the greatest outcomes of the study is that yes, differences exists between users but they are quantifiable and trends are clear to be seen.
How long before AIs will automatically adjust the EQ based on the material been listened to and your Facebook account?
Note that measurements for physical (anatomic, fitting etc.) variations is very much possible already (otoacoustic emissions and self/adaptive EQing products)
3. based on your 'casual observation' of what you think is...
4. one particular listener's preference which applies specifically to them...
5. in their sighted, non-level matched conditions, so even if correctly inferred by you, may not match with true blinded, level-matched preference...
I don't see the point here. I just stated that to my casual (note the importance of this word) observation based on quite a few EQ he devised for Headphones + loudspeakers Amirm seems to prefer;
@amirm seems to be +3dB@ LF -1.5dB at HF. That's all. He has on several occasions acknowledged this to me directly on the forum. What's the big deal?
According to you he may or may not know what he prefers but he seems to be rather consistently, not perfectly, exhibiting the same trend which is the crucial point to observe for a casual reader of Amirm subjective reviews.
6. a target which @Robbo99999 has pointed out isn't consistently matched in the same way to as your Harman target EQ, as is the case with the EQ for this headphone.
I have clearly explained why I did that for the first EQ. Amirm who listened to the HP did not feel the need to further EQ.
I took that onboard.
I very seldom do that, and it seems I should not have done it here for some reasons that escape me.
As you pointed there are other resources on the web, everyone is free to use them and prefer them to mine.
The second Flat@HF iteration is a much closer to the target after
@Robbo99999 "kindly" pointed that out.
Part of my concern is that if the EQ users are reading and buying their gear based on ASR reviews, they will have amps with plenty of headroom. Which is good, but even by loosing 12/15dB of sensitivity through EQ their system might still have more that enough power to damage their headphones.
So you say I should disregard that? well I usually do but do provide warnings...
7. All this accompanied by not scientifically valid adjusted 'scores' with this new target that have no basis in research that specifically uses the Harman target in the algorithm's calculation of the predicted preference score. You can't just change the target in the algorithm and expect that to be valid.
That's is the only point that I would gladly concede.
The weights in the score with a different target may or may be identical amongst other possible differences.
However the -1.5dB target scores 99.8 against the default target so yes different, lower (which should satisfy you) but not miles away.
That would need to be compared with the actual listening score error bar and other things to be judged.
I have observed that in some cases the EQ devised by Amirm would actually decrease the score while he reported a significant improvement.
The one simple way to get the score consistent with his rankings was to modify the target curve according to the trend that I have casually observed.
Assuming that Amirm is indeed roughly +3dB@ LF -1.5dB at HF his actual preferred target is no longer the default curve so the scores/rankings derived from it are not completely accurate for him. It makes more sense to me to swap the target curve than modifying the weights of the metric.
But that is up for debate...
Nevertheless, the great news is that the model is good enough to detect the score vs listening ranking inconsistency with the default target.
This result tends to validate Amirm not been right on the default target curve BTW.
Plus while being rough, my fix reestablishes the correct score vs listening ranking consistency with the modified target.
The actual numbers not being that relevant in this case, I'll admit to that.
That's the way science works, 1. establish a model/theory 2. compare it to observations 3. adjust model/theory to fit the observations
I will definitively make it clearer that the scores are not comparable but again it is clearly stated that the target curves are different.
I can still however target something different but calculate the the actual score which may be consistent for some users but not for others.
In the AKG case I only modified it at HF to try to prevent over-boosting.
An approach such as Oratory's is better here - EQ to the Harman target, but carefully curate the bands to include some broad filters such that regions like bass and treble can be fine-tuned according to preference / headphone unit / listener anatomy / mixing or mastering variation.
As you pointed there are other resources on the web, everyone is free to use them and prefer them to mine.
But if you are referring to the magic band #3
Oratory's:
Adjust gain of band 1 to preference (bass)
Adjust gain of band 2 to preference (warmth)
Adjust gain of band 3 to preference (tonal accuracy)
Adjust gain of band 7 to preference (treble detail)
My answer it still the same:
I agree with what oratory wrote. My point is this: when all DT770-250 differ from the average and you create an 'exact' EQ (multiple bands) for that average then probably none if the DT770-250 will sound 'accurate' as all of them differ audibly. They will all sound equally different but just...
www.audiosciencereview.com
What is stated here is quite tautologic in essence: if you change the system will sound different.
In particular band 3, I though we were looking for accuracy so I don't see how it can have different settings.
This band it is either accurate or not.... That is definitively not the apanage of a single band.
One more line of warning I put in my all my headsets EQs:
- The EQs are starting point and may require tuning (certainly at LF).
I cannot predict the users taste so up to them to tweak. Expect a learning curve though.
I sometimes use Amrim's preference (documented as more bass and less HF) and modify the Harman curve to show that for him his EQ make sense. but that's all.
As a general comment my algorithm does not simply optimize for the curve fitting nor the score.
There is way more going on in the fitness function of the evolutionary algorithm and it is a WIP and it has been for years.
I welcome any feedback, good or bad, in order to improve it.
Oratory know what they are doing, there is no denying that, but it does not mean that others can't do a similar job.
We'd better leave it at that.
Furthermore if ANY rational or otherwise deviation from the default Harman curve is "arbitrary unscientific" then Oratory should not provide any facilities to encourage them either. I suspect we both understand the studies, just replace his "preference" by my "taste" we are advocating the same thing. The difference is that I went one step further (but not better) by directly proposing (the first) two EQs that for one takes into account Amirm's input and for the second (-1.5dB@HF) fits a different target curve to try to mitigate the boost at HF while been reasonably close to the default target.
.
