Don't f***ing drag me into this.
You've dragged yourself into this
metrological mess (well, with funding help from your
sponsor headphones.com of course). As even you yourself admit in that video, the 5128 is "very, very inconsistent", and without this precision, (supposed) higher accuracy is practically moot.
But it is preference though. In the AE/OE research, respondents, within a controlled environment, were told to adjust bass and slope levels to their taste. The average of the adjustments then created the Harman AE/OE target.
I will remind people of my
previous comment that you ignored. To make sure you actually read and understand it this time, I will here quote and explain Sean's comments I linked to there on Harman's use of preference as an accurate proxy for perceived neutrality:
Toole spent 10 years having listeners rate loudspeakers based on perceived fidelity/neutrality. When we switched to preference, the loudspeakers ratings didn’t suddenly change. There is a high correlation between fidelity/neutrality/ preference.
Our headphone targets do not deviate significantly above 200 Hz from a anechoically flat speaker measured in our reference room at the DRP. For the AE/OE target it’s within 2 dB of the bass of the in-room speaker target. For the IE target it’s higher, but there are data to support it needs to be higher to be perceived as equivalent.
The switch to preference was a matter of improving methodology, as 'neutral' has become distorted in the minds of some people (not least reviewers, you being one) from its original meaning (with its root in the Latin 'neuter', meaning neither), which is literally 'neither too much, nor too little (of anything)', which in turn logically entails most preferred when the question is posed and understood correctly as perceived neutral. Moving from a semantically loaded (for some) term like neutral, to preference which has a universal common unidimensional meaning that's impossible to misinterpret, just removes potential confounding of the results due to inconsistencies in interpretation of the former. It's exactly the same with Harman's blind MOA tests; when participants reach their
preferred bass/treble levels, that is perceived
neutral to them i.e.
neither too much nor too little of either. And these average preferred (perceived neutral) levels were close to those of loudspeakers based on an 'objectively neutral' speaker's (i.e. anechoically flat) response in a good room, the difference (mostly in the bass) due to perceptual differences (among them the SLD effect, see
Theile 1986) between hearing sound from speakers and headphones. This is the whole point of Harman's research, to arrive at standard targets for both speakers and headphones that share this perceived neutrality among the vast majority of listeners. And they achieved that. The segmentation paper often erroneously cited as evidence for a multiplicity of preference and targets is
anything but.
What I did criticise, however, is the Harman In-Ear research
Oh it was far more than criticism. Again a reminder:
All I need to do is to cite is the title of research that went into developing IE 2017 2016...Respondents in this study (if you can even call it that)
Calling Harman IE "garbage"
I wonder if you'd dare say those things to Sean's face the next time you see him? Try asking if you made him uncomfortable after that. Then there's the blatant misinformation about the target, which you were made fully aware of but still chose to disseminate further without correction on your channel. None of this are the words and actions of someone just offering a fair criticism of Harman's research as you're now trying to rewrite history by claiming. Even once corrected your argument is moot, because we have independent
corroboration of Harman IE target preference and the predictive preference algorithm based on it. And then there's the hypocrisy of vilifying this scientific research, yet making sweeping claims like 'a lot of people don't like Harman IE' that you back up with...anecdotal internet polls, and not a shred of valid controlled blind preference evidence as Harman have backing their target. It's ridiculous.
I don't think I had endorsed or promoted any sort of specific 5128 target to Sean. The opposite really; I asked if it was a good idea to follow his example and allow for user-specified target curves just like how respondents had that choice in the Harman AE/OE research (and emphasis - not the IE research). Instead of promoting a "one true curve", simply acknowledge that user preference exists and have users specify their own parameters, and going even further allowing them to EQ their headphones/IEMs to their preference of slope and bass emphasis with the tool's built in AutoEQ function.
This is absolutely
not 'just like in Harman's research'. That was done via
controlled, level-matched, double-blind listening. Few if any of your followers will be doing this while playing around on your graph tool with made-up target curves under heavy influence of innumerable cognitive biases (like 'Harman IE shouty!' propaganda being relentlessly shoved down their throats). It's just complete pseudoscience. And once again, you have entirely misinterpreted the segmentation paper and are fallaciously using that to support a pluralist choose-your-own-target nonsense narrative (
in concert with headphones.com) that runs counter to the whole basis of Harman's research with its foundation of ending the circle of confusion with a (at least de facto) standardized target.
