Originally Posted by penngray
Perfect as in pure measurements?
Perfect as in what you want?
I tend to think perfection is a waste of time if its about what someone wants. All I care about is that does the speaker beat all other speakers in the important measurements. CSD, IMD, Polar, THD etc
I used to be a hardcore / textbook objectivist, and I would scour the audiophile mags / web sites for speakers with measurements that were as close to theoretical perfection as possible. I’ve read much of what Sean Olive and Floyd Toole has written about objective speaker performance and how they tie to listener preferences, so I went looking for speakers with flat on-axis FR (+/-1dB), well behaved off-axis FR, low cabinet resonances, no ringing in the spectral decay, low distortion, etc. My thinking was that if I get speakers that were as accurate as possible, I would get a more realistic sound, and that I would get just what was on the recording. If something didn’t sound pleasing through said speakers, then the problem could be clearly blamed on the recording.
I’ve owned three such speakers in my lifetime (Aerial Acoustics 7B, Revel Studio 2, and KEF Reference 203s), and despite all of them having very similar performance on paper, they all had distinct character that could be heard across a variety of recordings. With all due respect to the owners of the following speakers, these were my impressions. The Aerials sounded a bit too dark and boxy, and I was never convinced that I was ever doing more than listening to a hi-fi system—no sense of the performer in the room. The Revels had a clinical quality to them, and they were also unable to suspend my disbelief. It was always quite apparent that I was listening to a recording of a performance through those speakers. While they sounded neutral they never sounded real or natural. The KEFs gave me moments of transparency, but they had some sort of coloration in the midrange that made voices sound nasal, sharp and thin. Forgive my use of subjective terms to describe the character of these speakers, but since what I heard does not line up with the measurements, it’s all I have.
I still think that measurements are a good starting point and that they can give an audio enthusiast a general indication of performance, but either there are things that affect sound quality that aren’t shown in the measurements available in Stereophile, Soundstage, Home Theater Mag, and Ultimate AV, or despite my efforts, I don’t know how to fully interpret them. So if I were speaker shopping today, I would still reference the published measurements on a set of speakers before listening to them to be aware of any potential issues, but I will no longer rule out a set of speakers because the measurements aren't close enough to perfect for me. My current speakers have on-axis FR that is +/- 3dB, and a suck-out in the crossover region if you measure below the tweeter in the vertical axis, yet to me they sound more natural and real than any of the more accurate speakers I’ve owned.
But lets just say for the sake of argument, that there existed today a set of speakers that were theoretically perfect. A sphere* that expanded and contracted instantaneously (perfect impulse response), with 1Hz – 100KHz FR that was +/- 0dB in all directions, zero distortion, no cabinet resonances (because there would be no cabinet), and phase coherent across all frequencies (because there would be only one “driver” and no crossover).
If someone likes to listen Brandy or Christina Aguilera (my wife does) and is looking for a balanced sound, they will run far, far away from these speakers. The problem is, the vast majority of consumers don’t buy speakers off of how well they measure-- they listen. So even if they were standing in front of the perfect speaker, most wouldn’t know it, nor would they care if their music doesn’t sound good through them.
*Note – it may have to be a semi-sphere as it is often debated as to whether omni-directional sound is the goal we are after.