Technical Talk--Can Speaker Performance be Measured

In three words, yes and no. Now let me expand on that oversimplified answer. Many aspects of a loudspeaker’s performance can be measured, but we are concerned, first and foremost, with somehow de fining the sound quality of a speaker system (normally, a pair of speakers positioned somewhat arbitrarily in an undefined room). This, after all, is a major goal of the speaker reviews in STEREO REVIEW and other audio publications.

Let us approach this problem from two directions: (1) the acoustical stimulus of the listener’s eardrums by the sound field created in a room by the speakers, and (2) the electrical signal presented to the speakers’ input terminals by an amplifier.

The characteristics of the input signal are (or can be) known with considerable accuracy and can be taken as the reference to which the acoustical stimulus is compared in order to establish how faithfully the input signal has been translated into an acoustic pres sure wave entering the listener’s ear. One might assume that if the two waveforms (acoustic and electrical) are identical, then the transfer of the electrical input signal to the listener’s brain (or at least to one of its entrance ports) has been accomplished with a reasonable degree of accuracy. Alas, things are not that simple.

For one thing, in order to measure the sound pressure at the eardrum, it is necessary to insert a tiny probe micro phone into the subject’s ear. This is hardly a normal listening condition, although the procedure is used in psychoacoustic research (usually to mea sure the subject’s hearing acuity rather than the quality of a loudspeaker).

The usual alternative way of dealing with this matter is to use a calibrated measurement microphone placed approximately where a listener’s ear might be located in an actual hi-fi installation. Since reflections of the sound from the boundaries of the room, as well as the standing waves created in the room by those reflected signals, can have a profound effect on the result of such a measurement, it is often made in an anechoic chamber (a room designed to absorb sound on all its inner surfaces, thereby eliminating or reducing reflected energy).

There are several problems in the use of an anechoic chamber for measurement of loudspeaker sound quality, although it can be invaluable for determining the characteristics of individual drivers. In order to remain echo-free down to low-bass (or even mid-bass) frequencies, the chamber must be very large, and consequently very expensive. Although many loud speaker manufacturers have anechoic chambers, few of these are big enough for accurate measurement of the bass output of even a moderate-price speaker. Fortunately, there are other methods of realizing many of the benefits of anechoic measurement without the use of an anechoic chamber. In general, these involve a mathematical computation process known as the fast Fourier transform (FFT).

There is a more serious difficulty in the use of anechoic test chambers for evaluating speaker sound quality, as opposed to the frequency response of specific drivers or combinations of drivers. The problem is that we do not listen to high-fidelity music systems in anechoic chambers. The sound is, to put it mildly, unsatisfactory. The acoustic properties of the listening room can have as much to do with sound quality as the actual performance of the speakers themselves.

Although this linkage of room and loudspeaker performance makes the prediction of a speaker’s sound in any particular room rather uncertain, there are still ways to make measurements in an ordinary room that can be useful for a rough overall assessment of a speaker’s capabilities. And these methods are much less expensive (and more economical of space) than building an anechoic chamber.

Unfortunately, there are no meaningful test standards that enable measurements made by different people to be interpreted by others to predict the sound quality of a speaker. It is difficult enough for a tester to do that from his own measurements, let alone from someone else's.

Next month, I will describe in some detail how I test speakers and interpret those measurements, but I would like to close this column by commenting on a somewhat related problem posed by a reader, who wonders about the effect of the nonlinear frequency sensitivity of the human ear on loudspeaker com parison tests. He postulates two speakers, one (A) with a rising low-frequency response and the other (B) with a “flat” response. (Let’s not worry at this point about what a “flat” response from a speaker might be.)

Aware that a speaker comparison must be made at identical listening levels (also not as easy to achieve as some people think), this reader suspects that at low levels speaker. A would sound relatively flat, since its inherent response resembles a loudness-compensation curve. (Such a response curve, as supplied by the loudness control of an amplifier or receiver, is intended to compensate for the human loss of sensitivity to low frequencies at reduced listening levels.) On the other hand, speaker B would sound deficient in lows because of the same effect. At a higher listening level, how ever, the emphasized lows of speaker.

A would be audibly excessive, where as speaker B would presumably sound as “flat” as it measures. By extension, a speaker deficient in lows should sound more accurate at high volume levels.

The reader’s question is, “What is the correct listening level at which to compare speakers?” He suspects that, if comparisons are done in a dealer’s listening room, they should be done at levels similar to those one would use at home.

I think he is correct, although I must confess I have never given the matter much thought. I never give too much weight to any speaker demonstration I hear outside my own home. I have rarely found any speaker that sounded the same at a manufacturer’s or dealer’s demo as it does in my own listening room (it might be better, worse, or simply different). This may result in part from the relative volume-level differences, but I suspect it is largely a function of room acoustics or the specific locations of the speakers. A manufacturer’s demo, if it is done correctly, is usually preceded by careful positioning and orientation of the speakers, along with acoustic treatment of the room. The moral is, don’t buy any speakers unless the dealer will permit you to return them if they don’t sound satisfactory in your own listening room.

Source: Stereo Review (Jan. 1991) by Julian Hirsch