The Thick And The Dead:
Page 2 2) Listen with real music. A reviewer who bases his opinions of a speaker's performance on Jazz at the Pawnshop, the JVC Audio Symphony, or the Sheffield sampler CD should be tossed on the pyre incinerating all extant copies of said material. Aren't you, dear reader, more interested in how the device under review handles Mahler's 5th, "Vissi d'Arte," Woody Herman, Furtwangler, Ella, the Schumann E-flat Quintet, or Muddy Waters? I'll allow Marni Nixon, but not Amanda McBroom. You can have two Telarc bass drums and one slam of the garage door. That's it.

3) Worry about ends, not means. After a while, reviewers begin to give credence to whatever manufacturers are promoting this year. A current craze is the metal-dome tweeter, a device which, due to its rigidity and low moving mass, scores well on the kind of waterfall FFTs we all know how to perform. This is a case of a material tailored to a measurement, forgetting two cardinal rules for speaker transducers: their diaphragms exhibit a high degree of self-damping, and generate low amounts of diaphragm noise (the eigentones made by the material when set in motion). Metals of all sorts are lousy in both respects. True, the best metal-domes sound better than the worst soft-domes. Still, I would never have a metal tweeter in my speaker system as I cannot abide the tinfoil colorations. Damping the metal with soft-dome materials such as polyamide seems to give the worst of both worlds.

4) Know your software. In particular, know how the recording was made. Distantly miked recordings done with omnis are forgiving and sound great on speakers having a forward tonal balance, as well as many minimonitors. Close-miked cardioid and other directional mics yield recordings which are often harder to reproduce accurately and pleasingly, and this is not necessarily the fault of the recording. If a speaker only sounds good on the distantly miked stuff, or on xylophone solos, that doesn't necessarily mean it's a winner.

5) Have a reference system. This is a toughie; you actually need two---preferably three---preamps and power amps, and remember the differences between them. Most frequently, differences in amplifier tonal balance show up at the frequency extremes. A good speaker sounds less warm on the high-damping solid-state amp with the undersized supply, quite full on the tube gear, and in the middle on the good (I hesitate to say "neutral") electronics. The gear should stay constant for quite a while, maybe a year or so. Costly and inconvenient.

6) Don't be a closet designer. Which is why I'm unsuited for this work. Designers invariably hate gear not designed to their prejudices. I dislike dipoles, for example, and always find fault with them regardless of their sound. Of course, promoters like the guy who claims a -3dB point of 18Hz for his dipole are only grist for my mill. Yes, I can conceive of a measurement condition under which such a result might be obtained. No, that is no guarantee you will actually hear anything like true, extended, linear low bass from that speaker. Praising a dipole for its bass is like loving a Ferrari for its fuel economy.

7) Make useful measurements. The standard pressure-amplitude response curve which infests speaker reviews around the planet is often worse than useless, particularly when made "on axis" (where, oh where, is the "axis" of a planar or multidriver array system?). If you want to make measurements, make four simultaneously, and display the results one above the other: amplitude ("frequency response" and good luck), phase change with frequency, modulus of impedance, and THD with constant drive yielding 90dB/SPL at one meter for all audio frequencies. As far as I'm concerned, you can stop the test when THD surpasses 3%, a point at which the sinewave is quite visibly distorted. Abrupt phase-angle changes and phase change not linear with frequency indicate a disregard for phase coherency. Impedance humps more than double the mean value; or, appearing at more than one, max two, frequencies, show the designer didn't do his homework.

I suppose there is more I could go into: the reviewer should have an ability to differentiate high-Q behavior (such as engineered-in panel resonances) from real bottom end, or sense enough to operate with program sources of correct absolute polarity, or think to place speakers at the correct vertical height for listening (assuming the manufacturer knows what that is), etc. However, I hear my faithful ethnic sidekick/employee, Kato, yelling that he's run out of speaker-cone copolymer. Since we like to do everything here In-House, I must now hightail it out to the company oil well and get, er, cracking. "Kato, did you park Black Beauty under the gusher again?!"