How to Audition Speakers

(Originally published in: Stereo Review, Sept. 1998)

Surely there are few more subjective tasks than evaluating speakers solely by listening. One audiophile's meat is another one's poison, after all. But among experienced listeners - those who have spent hundreds of hours comparing speakers under controlled conditions and who agree on a set of criteria - there should be some common ground regarding methods and conclusions.

That's not to say we have to agree. If your notion of audio excellence runs to maximum bass-slam for dance parties, so be it. The advice, "Buy what you like," has the virtues of being direct, easily followed, and by no means wrong. But if you're prepared to accept "accuracy" as the loudspeaker ideal, you should find the following helpful in finding speakers that exemplify it.

Accuracy in this case can be defined as reproducing the nuances of tonality, dynamics, and ambience in the original recording with as little change as possible, over as broad a useful frequency range as possible, within the relevant constraints of cost and size. Of course, this definition is circular, because how can you know what the recording "sounds like" until you play it back on some speakers?

You can't - not any more than we can know just how closely the Mona Lisa resembles an actual sixteenth-century noblewoman. However, we know that Leonardo da Vinci was a skilled painter and accept his rendition as both an accurate representation and something more: a work of art. Similarly, if we choose an artful recording, we can assume that its playback has the potential to be a fair facsimile of the original performance. And even though recording equipment (like microphones) varies in inherent sound quality in almost precisely the same ways as speakers do, the essential qualitative differences among excellent recordings should be far smaller than those between an excellent and a mediocre speaker.

On the Job

There is no substitute for experience, so embrace every opportunity to listen to and compare speakers. And remember, the listening conditions are critical. Always use an excellent amplifier or receiver and a high-quality source component to minimize the variables in what you'll hear. But you can rest assured that the immediately audible differences even between similar speakers are likely to be worlds greater than those between properly operating amplifiers and disc players.

If you listen with an A/B setup that allows you to switch directly between two pairs (or suites) of speakers, proceed with caution. If you're using an A/B "switcher," make sure that the comparison is carefully balanced so that both candidates play at the same volume, within 0.5 dB. (This can be determined only with an accurate sound-level meter.) It is a psychoacoustic fact that if one speaker is more sensitive than the other, and thus plays even slightly louder, it will be perceived as "better."

Here's another tip: placement affects sound quality more than you might guess. Whenever possible, reverse the locations of speaker pairs you're comparing, even when they're side by side, and you'll hear just how much location matters. But don't rely exclusively on A/B comparisons. Also include some one-at-a-time listening sessions to get to know the sound of each speaker pair or grouping.

Obviously, the program material you use is very important. Choose demo discs not for their musical excellence, but for their recording accuracy and ability to highlight particular aspects of sound quality. There are many approaches to selecting evaluation discs. I usually rotate a few familiar tracks so that I can zero in on key sonic characteristics (see "Following the Tracks" on the next page). Not the least of this method's advantages is that you quickly become so thoroughly sick of these snippets that you cease to hear them as music and can focus on how accurately the music is reproduced.

A Listener's Checklist

And when all is said and done, for what, precisely, are you listening? You could ask this question of a dozen "experts" and get a dozen answers, so take any of them with a dose of salt, mine included.

1. TIMBRAL UNIFORMITY. By far the most important factor to me, this means even emphasis and true timbre ("tone color") over the broad range from bass to treble. In practice, timbral uniformity is impossible to judge as a whole, so I tend to break it into four subcategories.

o Vocal-range smoothness. Here I'm listening exclusively to voices, both singing and speaking, for freedom from midrange colorations - probably the most common speaker weaknesses. Most "errors" manifest themselves as tone-color constants that I can pick out in a variety of different voices. This can be a "honky" or nasal quality, a "cupped" tone (as if the words are sung or spoken through cupped hands), or a persistent raspiness or hollowness. I consider these the most serious of audible flaws because once I detect one I will hear it almost constantly, coloring the vocals of artists as disparate as, say, Janis Joplin and Gordon Lightfoot.

o Bass/mid-bass anomalies. These are usually simpler to evaluate. The common telltales are boominess or thinness (a tendency to accentuate or short-change certain pitches), or chestiness or hollowness (overbearing or anemic-sounding low male vocals).

o Low-treble smoothness. Massed orchestral strings are my usual test here. I listen for string tone that's unnaturally edgy or dry ("dead"-sounding) or overly syrupy. This is one of the toughest calls unless you have some experience listening to live string sections - and even the best strings can sound a touch harsh or steely depending on hall acoustics and playing style. However, keep an ear out for a consistently strident or metallic tone, as well as an unnaturally mellow or rich one. Try several recordings as these qualities can be artifacts of the recording itself.

o Tip-top treble. In a great deal of music little actually goes on above 13 kHz or so, and we hear much of what does occur up there more as "sparkle" or "air" than as musical notes. Listening to jazz or rock cymbals is one reasonably easy way to find treble hooks. High-hat rides - the "tick-tick-ta-tick-tick" that glues together so much jazz and rock - from naturally recorded discs make excellent test fodder. They contain clean, repetitive transients. Focus on the character of each individual "tick," noting signs of dullness (too little top-octave sound), "spittiness" (too much), and "smearing" (a sort of vague, lisping, "un-metallic" quality that real-life cymbals don't have).

2. IMAGING. Here I'm talking about spaciousness, soundstaging, depth, and the hundred other terms used to describe how well or poorly speakers conjure up the ambience of a real acoustic space. All in all, imaging may be the most variable speaker quality, and it's certainly the most difficult to evaluate. That's because even the best reproduced sound is not, in fact, very much like the real thing.

That said, I listen first for stability of the image. Does the apparent location of a lead vocalist or instrumental soloist "wander" from side to side, or front to back, as the music moves up and down in frequency? Is the soundstage arrayed across the full space between the two speakers, spread out beyond them on either side, or bunched up in the middle? Is there a natural sense of front-to-back depth to instruments and voices, without exaggerated (or repressed) reverberation?

All of these questions are tough, and there are no "absolute" answers. Every speaker distorts "reality" in one way or another, so these variations are a matter of taste as much as anything else. And different types of speakers have inherently different imaging characteristics. For instance, dipolar speakers, like most electrostatic or planar-magnetic designs, usually create an enhanced sense of depth and space, but at some sacrifice in the "precision" of imaging and the tight localization of the instrumental soundstage. Controlled-directivity designs, such as those using horns, and certain multiple-driver arrays tend to have just the opposite character, and the great majority of conventional two- and three-way speakers usually fall somewhere in between.

3. DYNAMICS. Virtually all speakers compress dynamics to some small degree. That is, beyond a certain volume, the speaker will not keep producing clean acoustic output equally over the full frequency range.

One form of dynamic limiting is obvious: audible distortions in which a driver, usually a woofer or midrange, produces buzzes, snaps, or pops on loud transients. Another, far subtler form is nonlinear response, in which the speaker fails to "keep" up with the input signal through some range, usually the deep bass. The result may be a progressive "brightening" at very loud volumes (the midrange/treble section can usually deliver a higher level of clean output than the bass section), or a loss of weight and impact on strong, wide-band transients.

But limiting factors are notoriously hard to judge in casual listening sessions. Can you be sure you're hearing a woofer "bottoming" and not the amp clipping? Speaker distortion and not buzzes or rattles from furniture, floorboards, or windows? Is a telltale brightening caused by speaker limitations and not by your ears' natural "distortion" at very high volume? Unless you're listening in a very familiar environment, you cannot be sure where to lay the fault.

4. BASS EXTENSION. It's a rare speaker that can really reproduce the full audio range, unrestricted, down to 20 or 25 Hz. Paradoxically, it's also a rare speaker that cannot produce relatively strong output from 80 Hz and up. So when we talk about bass extension, we mean the bottom two octaves in the ten-octave musical range - and the lower one, the 20- to 40-Hz octave, sees very little action from traditional musical instruments. (Big percussion, synthesizers, and soundtrack booms and rumbles are another story.)

Recordings with energy in the sub-40-Hz octave are relatively rare - big bass drums in classical music (like Copland's Fanfare for the Common Man) and ultra-low synth-bass lines in pop are your best bets. Unfortunately, it's very hard to judge the deep-bass response of a speaker until you get it home, because placement and room acoustics affect bass more than any other area of speaker performance.

Given all the above, I tend to evaluate speakers with one or two very familiar recordings that have solidly dynamic percussion. I listen for a sense of ease, "weight" (unfetterred deep bass), and realistic snap. But I qualify my judgements a great deal until I've had a chance to repeat the trial in my own studio. Only then am I confident enough to call a speaker's low end "good," "bad," "mediocre," or "absolutely fabulous."