I still sense a confusion of purpose in this discussion, at least when my name and the Harman tests come up.
Back in post 1043 I said:
"Yes, mostly because of a confusion of purpose. The double-blind, equal loudness multiple-loudspeaker evaluations were designed to maximize the detection of resonances, not to determine if the overall spectral balance flattered specific recordings - recordings do not have consistent spectral balance. Resonances are more easily heard in mono - discussed in Chapters 4 and 5. The elimination of resonances, results in smooth, flattish direct sound, and if directivity is also smoothly changing, high sound quality ratings. If such a spectral balance is not found to be enjoyable in a particular situation, it is then a challenge to decide what is responsible. Often a bass issue (30% of our overall sound quality rating) which is a small room acoustics problem that is often inadequately corrected. Excessive bass is easily confused with insufficient HF, and vice versa. As I have said, only trustworthy anechoic data will describe what the speaker is doing, and in-room measurements are needed to evaluate what the room is contributing. Spinoramas enable the variables to be sorted out and addressed."
All mono tests were comparisons of 3 or 4 loudspeakers designed to maximize the detection of resonances, the most audible problem in loudspeakers. Listeners rated the sound quality as if they were judging absolute "fidelity" to some conscious or subconscious criteria of what sounded best. However, because of the rapid randomized comparison method of the test, the unspoken question being asked of listeners was "what is the loudspeaker doing to the sound?". Positional substitution rendered the room a constant factor, and 30 s musical loops ensured that all loudspeakers were evaluated with the same monotonous test sounds. The dominant variable was the loudspeaker. Problems were quickly revealed, and as could be seen in measurements resonances were clearly the dominant problem. They were easily heard, and the highest rated loudspeakers exhibited the least evidence of resonances, which by definition corresponds to smooth flat frequency response in the sound which is perceptually dominant, the direct sound. It was quickly observed that the highest rated loudspeakers also had smooth off axis frequency responses. This has been true since the first tests I did in 1966.
The music was a variable, but only in the sense that not all music revealed all resonances, and certainly not all at the same time. Some music was much more revealing than others. Pink noise was the most revealing of all sounds. Solo voices and instruments were generally the least revealing - use these to demo flawed loudspeakers. Measurements reliably reveal all resonances, audible or not.
But, then comes the question that reviewers and consumers ask: how does the loudspeaker sound with recordings?. This is where the "reference axis" might matter. But, at this point we confront the "circle of confusion", because recordings are not standardized test signals. They were mixed and. mastered using loudspeakers in rooms that are not the same as yours or mine, not even considering how they might have been equalized or acoustically treated. There are no standards.
Differences in spectral balance inevitably occur and over the years playback electronics have included tone controls, occasional "tilt" controls and now user-accessible parametric equalization to compensate for these variations. Fussy listeners use them - they are not "fix and forget" controls. Fortunately, things today are far better than in years past - just listen to some of the old "classics" through modern, neutral, loudspeakers. Remixes have helped in some cases, but we learned to like them with flaws.
The discussion about which listening axis is most flattering to most recordings is academically interesting, perhaps, but in the real world a tweak of a tone control or simple EQ fixes many concerns - a different loudspeaker is not a requirement, so long as it is not burdened by resonances or gross frequency-dependent directivity problems.
Finally, humans adapt. We normalize, learn to ignore, some kinds of flaws in sounds we spend time with. Problems may be there, but we are not aware of them. This explains why so many flawed loudspeakers have been satisfying so many listeners for so long. The adaptation can be instantly broken if there is a comparison to a superior sound, which is what drove the scientific investigations, and the multiple-loudspeaker comparison method.
Overlayed on all of this are the soundstage and imaging illusions of stereo. Inevitably the room is involved and no two are alike, so debates arise. Again there are no standards and the principle determinant of these perceptual factors is the mix itself, not the playback apparatus.
Does this make sense? All these topics are elaborated on in the 4th edition of my book for those who might be interested. Learning to interpret spinorama curves is a huge asset in choosing a loudspeaker, and using it to its best advantage. In-room bass measurements and EQ complete the exercise. Relying on subjective opinions formed in the commonly available circumstances leads nowhere reliable - but it feeds forum discussions.
