Analytical or Musical Which way to go?

Drew, I agree. However, many audiophiles dislike tone controls. This I think is a shame because while they shouldn't be, recordings are all over the place, with some so hot that they're virtually unlistenable on a good system.

From a speaker manufacturer's perspective, the speaker has to sound good *as is,* because that's how it's demoed and that's how most people will judge it. Some manufacturers go for flat, some go for a downtilt, most have poor dispersion in the highs and, unfortunately, most records are mixed to speakers that have poor dispersion in the highs. Now that digital EQ is readily available, I don't think it much matters. Polar response is more important, because that can't be fixed with equalization.

I've always found that the best components are the most accurate ones, as measured by their ability to recreate something like an original performance.

"Analytical" components aren't always the most accurate in that respect. For example, a speaker that beams will likely sound more analytical, because it minimizes room reflections. But two channel stereo is missing the lateral reflections that are crucial to a sense of acoustical space. So a loudspeaker that illuminates the walls will if the speakers are a sufficient distance away sound less detailed and more spacious -- less analytical, but more accurate, by comparison to the original sound.

In the case of measurements, I'd distinguish between naive measurements and properly-interpreted ones. An example of a naive measurement would be the assumption that the frequency response of a speaker should be a horizontal line. Speakers that have that kind of response don't sound like the original performance with natural two channel stereo. They sound too bright. This effect has been known for years and is probably caused by the rolloff of high frequencies in the reverberant energy of a large acoustical space. Cardioid microphones and microphones that are too close don't pick up as much of this energy, and so the balance shifts and becomes too bright. It has to be compensated, either in the loudspeaker, or in the target curve of room EQ.

Again, the goal is an accurate reproduction of the performance. A flat speaker may measure well, but the total frequency response of the recording/reproducing chain will be wrong.

Gregadd, funny that you mentioned John Atkinson's article because I was about to mention it when you asked whether measurements can predict colorations. What I was going to mention was in Part 3 of the article, though. Some of his conclusions:

"What Makes a Good-Sounding Loudspeaker?

"Vance Dickason offers some discussion of this question, but the definitive answers are to be found in Floyd Toole's comprehensive 1986 papers. Nothing that I can conclude from my past eight years' work, at least when it comes to conventional forward-firing, moving-coil designs, is in serious conflict with his findings. As I wrote in 1991, 'The best-sounding loudspeakers, in my opinion, combine a flat on-axis midrange and treble with an absence of resonant colorations, a well-controlled high-frequency dispersion, excellent imaging precision, an optimally tuned bass, and also play loud and clean without obtrusive compression'."

And

"Most important, while measurements can tell you how a loudspeaker sounds, they can't tell you how good it is. If you carefully look at a complete set of measurements, you can actually work out a reasonably accurate prediction of how a loudspeaker will sound. However, the measured performance will not tell you if it's a good speaker or a great speaker, or if it's a good speaker or a rather boring-sounding speaker. To assess quality, the educated ear is still the only reliable judge."

http://www.stereophile.com/content/measuring-loudspeakers-part-three-page-9

The Harman research seems to correlate well with what audiophiles like, though. The main exceptions I can think of are planars, as John Atkinson pointed out in his article. Floyd Toole sheds some light on this in his book -- the Quads (57's?) tested much better in stereo than in mono. And Harman tests in mono. Their speaker positioner also doesn't substitute for the careful setup of an audiophile system, which is particularly crucial with dipoles.

One point that Olive and Toole make, and it's one with which I agree strongly, is that speaker preferences are not solely a matter of taste. In blind tests, subjects with normal hearing routinely pick the speaker that is most accurate. I don't find that very surprising. Of course, we also "choose our poison" to some extent, depending on our listening material and levels and what we value most in reproduced sound. But the notion that people prefer inaccuracy doesn't seem to be true.

I'd draw a distinction, though, between picture-perfect response and accuracy with real-world material. If, say, pop recordings are hyped in the highs, as many are, you're likely to want a speaker that compensates for that.

""How do we know which componet is colored? Or if the source is colored? It has been suggested to me that EQ negatively alters the tonal balance of a speaker."

Always a problem. I do three things:

- Use well-recorded reference disks for system evaluation and setup. That generally means naturally miked, purist recordings of acoustical music, made in a good hall. Multimiked commercial recordings will generally be bright or screechy by comparison, and don't image well.

- Use measurements. They don't tell you everything, but they do tell you which components are flat or not, since aside from the speakers, they should all be a straight line. But usually these days, unless you're using a high output impedance amp, it's the speakers that are the main culprits, followed by the cartridge. I'm not a big fan of using one colored component to compensate for another, though in that I'm probably in the minority these days. That's because I think there are cheaper and more effective ways to do the same thing, namely EQ.

- Just listen, because in the end, none of it really matters, it's what you hear with the recordings you listen to that counts.

BTW, I think those who are criticizing EQ are thinking of the old analog jobbies. They rarely put the correction quite where you needed it and they tended to introduce other colorations as well. DSP-based EQ is much more effective. However, unless you have a lot of time or are perhaps making a copy of a favorite recording, when it comes to correcting individual recordings I think most of us have to settle for a rather coarse adjustment of the sort that's best suited to really unlistenable recordings. So I'd begin with a few target curves -- maybe hot pop, large ensemble classical, small ensemble classical, all in series with your room and speaker correction, which should be constant. Then I'd use manual correction to make the really offensive recordings listenable.