What Do Measurements Really Tell Us?

Modern loudspeaker measurements can produce an astonishing amount of information. A computer can generate beautifully detailed graphs of frequency response, distortion, impedance, phase, decay, and other characteristics within seconds. The danger is assuming that more data automatically produces more understanding.

A measurement is useful only when we understand what it represents. No single graph tells us whether a loudspeaker is good, and no collection of graphs can listen to music. Measurements are valuable because they reveal physical behavior that may be difficult or impossible to identify reliably by ear alone. The important question is not, “Does this loudspeaker measure well?” but, “What do these measurements tell us about how this loudspeaker behaves?”

Frequency Response

Frequency response is probably the most familiar loudspeaker measurement. It shows how strongly the loudspeaker reproduces different frequencies under specified conditions. A reasonably smooth response is desirable because large peaks can emphasize parts of the music while deep depressions can obscure information.

A flat frequency-response curve, however, does not guarantee a good loudspeaker. Two speakers can have remarkably similar response curves while sounding quite different because frequency response does not fully reveal stored energy, distortion, directivity, or transient behavior. The shape of an irregularity also matters. A broad, gentle variation may be relatively unobtrusive, while a narrow resonance of similar amplitude can repeatedly announce itself whenever music excites it. Frequency response is therefore important evidence, but it is not the verdict.

Off-Axis Response

A loudspeaker does not radiate sound only toward the person sitting directly in front of it. Much of its acoustic output travels toward the walls, floor, ceiling, and furnishings before being reflected toward the listener. Those reflections contain the loudspeaker’s off-axis response.

If the loudspeaker changes character abruptly as the listening angle changes, the reflected sound may have a substantially different tonal balance from the direct sound. The ear eventually receives both and combines them into one impression. A loudspeaker with an exceptionally smooth response directly in front of it can therefore sound less natural in a room than one whose forward response is slightly less perfect but whose radiation changes smoothly with angle. Off-axis measurements are among the most revealing tools available. They tell us whether the loudspeaker behaves sensibly as an acoustic radiator rather than merely at one convenient microphone position.

Impedance

A loudspeaker advertised as “8 ohms” is not actually an 8-ohm load at every frequency. Its impedance rises and falls as its drivers, cabinet, and crossover interact. An impedance measurement tells us whether the loudspeaker presents a reasonable electrical load to the amplifier, but it can reveal much more. Resonances often appear in the impedance curve. Changes in enclosure damping and loading can become visible there, as can some crossover interactions and characteristics of the driver’s motor system. Impedance is therefore not merely a number printed beside the speaker terminals. Properly interpreted, it becomes a diagnostic tool that can reveal what is happening mechanically and electrically inside the system.

Distortion

Frequency response tells us how much output exists at particular frequencies. Distortion measurements help reveal sound that the loudspeaker creates which was not present in the original signal. As a driver is asked to move farther or operate closer to its limits, its suspension, magnetic system, and other components become less linear. Harmonic distortion may increase, and different frequencies reproduced simultaneously may interact to produce intermodulation distortion. These effects often become much more significant as playback level rises. The objective is not simply to chase the smallest possible percentage. Human hearing is more sensitive to some kinds of distortion than others, and the same numerical amount can have very different consequences at different frequencies. Distortion measurements are most useful when they help identify the physical mechanism causing audible degradation.

What Happens in Time?

Music does not consist of continuous laboratory tones. Sounds begin, change, and stop, which means the behavior of a loudspeaker in time matters. Impulse response, decay measurements, group delay, and measurements of stored energy can reveal behavior that an ordinary frequency-response graph may conceal. A resonance, for example, may not appear particularly alarming in amplitude yet continue producing sound after the original stimulus has stopped. That lingering energy can blur transients and add coloration to music. Two loudspeakers with similar frequency-response curves may therefore behave quite differently when reproducing rapidly changing musical events. Time-domain measurements help show us why.

What Happens When It Gets Loud?

A loudspeaker measured quietly may behave beautifully while becoming a very different system at realistic listening levels. Voice-coil heating, suspension nonlinearity, excursion limits, and thermal compression often become important only when significant power is applied. Measurements at increasing levels reveal whether the loudspeaker maintains its frequency response, whether distortion rises gradually or suddenly, and whether increased electrical input continues producing a corresponding increase in acoustic output. A loudspeaker that compresses dynamically may survive high power without obvious failure while still losing the contrasts that make music sound alive. This information is generally more useful than a large “maximum power” number on a specification sheet.

What matters is not merely whether the loudspeaker survives the power. It is whether it continues behaving properly while using it.

Measurements Reveal Causes

This is perhaps their greatest value. If a loudspeaker sounds aggressive, listening alone may tell us that something is wrong. Measurements can help determine whether the cause is a response peak, breakup, distortion, stored energy, irregular directivity, or something else entirely. If the stereo image becomes unstable through the crossover region, measurements of phase, summation, and radiation pattern can help reveal why. Conversely, a graph may contain an unusual feature that appears alarming until further investigation shows that it has little audible consequence. A measurement must always be interpreted according to the physical mechanism producing it and the conditions under which it was obtained.

Measurements describe reality. Engineering requires understanding what that description means.

No Graph Contains Music

Some questions test instruments cannot answer directly. A microphone and computer cannot decide whether a piano retains believable weight and attack, whether a voice sounds naturally human, whether a stereo image remains convincing throughout a performance, or whether several hours of listening become tiring.

Those questions require listening because the purpose of the loudspeaker is not to produce excellent graphs. Its purpose is to reproduce music. The most useful measurements, therefore, do not replace listening. They make listening more informative by helping us understand why we hear what we hear.

Measurements tell us what the loudspeaker is doing. Listening tells us whether what it is doing serves the music.