I agree with many of your points
@b1daly. This is the one point I think needs to be qualified though.
Yes, it’s true that the concept of “accuracy” is fraught given the loudspeaker reproduces a two-dimensional signal (frequency and amplitude) in 3D space.
However, in my view, the only metric that this complicates is that of directivity. All other metrics can be referred back to the signal without complication (examples: nonlinear distortion, phase distortion, spectral decay, etc).
In other words, accurate reproduction of the signal dictates that the speaker should produce a flat axial frequency response with as little nonlinear distortion, phase distortion, spectral decay, etc, as possible - or at least to keep these deviations from the input signal below or as close as possible to thresholds of audibility.
What it doesn’t dictate is what the speaker’s polar response should be. That metric alone is one about which I think various competing arguments can be made, possibly with reference to listener preference.
Do you believe that these other metrics are also problematised by the fact that a loudspeaker reproduces the signal in 3D space, or are we more or less on the same page?
I definitely accept at face value the implications of the research, as it's being reflected back to me through various secondary sources. (I am going to read Mr. Toole's reference one of these days.)
I'm mostly trying to analyze my surprise at how this research differs from my own limited subjective experiences. I'm pretty confident that there is an objective element to what I am perceiving. But I could be mistaken. Maybe it's not as stable as I think. It could be that I am a weirdo who actually has different preferences than most people. (In discussions I've seen this argued as being a myth, but that is based on a statistical sample. If a person really has repeatable preferences that are stable, then that's all there's to it.)
From what I'm gathering, people's preferences have consistency across different listening environments, based on the brain's ability to distinguish direct and reflected sound. (This blows my mind, but we really do have amazing filtering going on in our brain all the time.)
But maybe the polar pattern response preferences are more affected by room treatment than the direct? Simply put, maybe a di-pole design has a different optimum listening environment from a conventional box speaker? Some speakers seem to be more sensitive to placement, is this a quantified effect in the preference testing?
If I had to put my finger on one aspect of the "accurate" bi-amped studio monitors that I find disturbing, is that the cabinets are often so dense, damped, and unboxy. One issue is that this exposes weaknesses in the drivers, so if there is an annoying sound in the tweeter, for example, it's plain as day. There are some big problems with commercial recordings being way overemphasized in the high frequencies, with not only signal but nasty distortion. Some tweeters seem to handle this better than others. (Out of the 30 odd sets of speakers I have, I have an old Polk Monitor 5A, and it has that original Peerless tweeter with the rectangular plate. I bought a Chinese made reissue, and that tweeter is the most forgiving I have ever heard for high-frequency distortion. The speaker sounds good, not great, but it has an amazing consistency in that almost everything I play through it sounds pretty good, compared to some more "detailed" sounding speakers where certain recordings sound shrill, the tweeters just can't handle the distortion in a flattering way.)
A lot of people like this "boxless" sound, but I find it uncanny, lacking in "solidity". My pet theory is that sounds don't just appear in space, they come from objects. On such a system a listener is presented with an experience where sounds that should be "coming from somewhere" don't appear to be coming from any discernable object. The sound is "disembodied." This is where I think vision must play some role in the listening experience simply because it's part of location sensory system.
One thing I don't have any perspective on is the relationship between frequency response and SPL from the speaker. Is this reflected in the distortion measurements like one finds here?
https://www.soundstagenetwork.com/measurements/energy_rc_10/
My wondering is along the lines of cabinet resonation, which to my ear seems to increase its contribution to the sound the louder the speaker is playing. This would be reflected in the 3D measurements if they tested at different power levels.
I think among our inbuilt pattern recognition are filters related to understanding the relationships between sound energy levels and things like physical resonance and distortion. A vibrating box has a very distinct sound, and I've always felt like it was part of the overall presentation.
I listen to music at very low levels, rarely exceeding 90db spl, usually averaging about 82-83 db. This presents some challenges. The loudness curve of the old days was an attempt to help this type of listening, which involved boosting bass and treble. It can work well sometimes.
Rock music presents a special problem for reproduction...unless you are young and crazy and blast the music. Audio producers put a lot of work crafting a signal that reads as "loud" when played at any volume. This effort is more or less successful, and it's based on manipulating the ambiance and timbres contained to "fake" those cues. This is very hard to do convincingly with DSP...actually getting loud objects going, with a lot of sounds and electrical energy input, will create very complex sound signatures, and when captured these can be incorporated to create psychoacoustic effects.
If you have an actual box, that vibrates, driven into distortion even, this can "help" in the creation of having a "loud" signal. I find that small computer type speakers can really help "glue" my own mixes together, giving a kind of natural, tough sound, because the amp overdrives, the whole cabinet resonates profoundly, and I know that they are really being "energized." Unfortunately, such speakers fail to communicate many other important aspects of a mix, and cannot be relied upon
I see this all over...people drive their playback systems into distortion. With rap and other electronically/digitally created music, the intended effect is only fully expressed on specific playback systems. A lot of people have these crazy subwoofer systems in their cars, they crank the shit out of them, and the very low frequencies found in rap music come "vibrantly to life" as they are actually vibrating the whole vehicle! Same thing with electronic dance music. It is intended to be played on very loud club systems, with a lot of bass, rendering a mix that sounds pretty flat on a decent system at a low volume, into a very exciting experience.
Anyhow, I'm brainstorming for myself along two tangents. I worked full time as a recording engineer for about 20 years, and I didn't even have a home stereo, because I couldn't listen to more sound. Now I'm only part-time, and I'm rediscovering home audio! And my home systems kick the ass of my studio systems in terms of my own subjective experience. (This might be because of other associations, work related stress, overly damped rooms).
I'm sick of working on systems that I think "sound bad" and I am looking for improved sound at home. Having mixed success so far, but I think it will try to find an affordable set of "well measuring" speakers like the Revel line to see what I think.
I see no reason why all sound characteristics of a speaker couldn't be measured, but maybe there are parameters that are quite in resolution yet.
I also think subjective experience is indeed the ultimate judge of an audio system. The foundation of the Toole research are such experiences, it just happens that group preference aligns with notions of "accuracy" in the electrical domain. It seems entirely plausible to me that processing devices, whether speakers or otherwise, could be created that would decrease accuracy and increase subjective listening pleasure.
For example, implementing frequency dependent dynamic limiting I think could be very helpful in "normalizing" the listening experience across varied sources. This might not show up clearly in controlled study, but something that is experienced over a longer time. (I think Denon is actually doing this in their AVRs). Or a magic distortion algorithm could, on average, make systems "sound better." The problem with such approaches so far, is that they are too dependent on the program, but my gut feeling is that we are going to get there.