I think in order to improve on the Harman Curve then another study would have to be done that was even more representative of the population, and maybe based on an initial Target Curve from a more anatomically accurate average mannequin and ear..........if that's possible, I suppose it is, but someone would have to determine what that would be and have it manufactured, and then carry out the study again. But all of that could be eclipsed in the future by personalised Target Curves & DSP based on head/face/ear scans perhaps to simulate accurately what Impulcifier and Smyth Realiser currently achieve, but with a lot greater convenience/speed and lower cost. I think the second option is the most likely, I'd be kinda surprised if Harman or anyone else redo the study and publically release the research and target curves again based on mannequin heads/ears. So I reckon we have Harman until the second option I was talking about becomes mainstream or properly workable...... (re good/fantastic results....speaker simulation in headphones).....if it ever does.
I wonder whether this discussion doesn't now necessitate a separate thread, lol.
Arriving at a "more anatomically accurave manequin and ear" could as well be better as to prepare different manequins for different population variants and then selling differently tuned models for a specific demographic. I believe that investing in easily available in-ear mics (perhaps even exposing the calibration mics in IEMs to be used independently by software) would be better in a longer timespan. Remember that there was Nuraloop with dynamic ear canal measurements for tweaks and also some developments. Simply put, feedback systems w/ dynamic calibration are the only way to go. Static calibration & tuning should go away.
As for the Impulficer, static speakers emulation is the best we can have without physical head tracking module, as I believe it is an integral part of Smyth's Realizer. Seems we're stuck paying a few thousand bucks to perceive 3D sound on headphones. Sad state of affairs...
... at least until someone works out on how to integrate dynamic mixing of sound into Jaakko's Impulcifer with some kind of freeware TrackIR-compatible simulation (like Freetrack), because that is essentially a well-known head-tracker for webcams (even cheap consumer ones) based on an IR filter and a really cheap DIY IR-based headset clip-on or a cap with diodes. I've used a DIY made by my dad to play around with Flight Simulator 2004 like ten years ago and it DOES work really well. Would sound reasonable to consider that Smyth is getting HRTFs of various angles for all of the speakers and then working with these. I don't know whether it would be easy to accomplish, but modern video games have very considerable soundspace simulation models. After all, it might be a matter of convolving channels w/ proper HRTF & then downmixing to stereo, but hey - I'm not a matter expert here.
I think it's still hard to say what target is "right" and "truthful" because a pair of headphones does not solely rely just on FR.
I think about it this way: there are headphones perceived as great and "natural-sounding" by both amateurs and pros, musicians and what-else-you-name-it but the thing is between those headphones, there are both non-Harman compliant and Harman-compliant ones so I fail to see how it's just the FR target dictating everything else.
Take what I write with a pinch of salt, since I'm not a certified sounds expert, nor a Masters/PhD in the matter.
The sound
does depend mostly (and very importantly) on the FR, since it is the direct and complete (many other factors can be inferred from FR) measure of linear performance of a device. The working theory around ASR is that high-Q (narrow in frequency) high-energy peaks might be the reason of varying "details-resolving" performance between various headphones. Masking phenomena are known for like what, over 30 years now? That's why I brought upon the FR's smoothness.
THD & group delay are measures of non-linear performance aspects of headphones. THD is already low in modern headphones. THD was explicitly neglected in Harman's limitations of 2013 study. Fit issues end up being FR & THD issues. Also linking a study from 2014 on
correlation of non-linear distortion vs listener preference. I didn't read that one before, but it's interesting to see that two-tone IMD on headphones can go up to 20%!
When bringing pro-audio market into the equation, remember that they *might* consider just buying anything that doesn't break after dropping it 2 meters high, the rest can be fixed in EQ to some extent if you need, as digital EQ is available everywhere. For monitoring, EQ-ing is very often not necessary (given that monitoring is a relatively narrow-band exercise) and mixing is done on speakers anyways, thus the context of recommendations & reviews is EXTREMELY important.
Whether impact of earpad size, depth & distance from ears has been extensively assessed in studies would be a good thing to research, as some have brought upon how HD800's have legendary soundstaging performance might be connected with their unique and unusual way of fitting the ears - this is the pair used in Smyth Realizer demos, maybe not without reason... Also, a disregarded but VERY important parameter is the L/R matching quality - very small deviations can cause sound localization issues (see
https://www.mdpi.com/2076-3417/6/4/117); this work also brings previous works which connect FR with proper front/back spatial perception (see section 6.2, paragraph 3 on "front-back confusion").
Theory I subscribe to, which I don't remember seeing in full form (probably personal compilation of some papers), is that mismatch between one's HRTF and headphones FR is one of the crucial factors on how a pair will be perceived. Get a high mismatch between the headphones vs. HRTF of your daily life and see the disaster happening. Seems that VR-related studies like
this one touch on this subject. Using headphones with a "wrong" FR should be similar to using a non-personalized HRTF (from a different person).