The Color of Sound

hearing the room

“The room is the most important component in your audio system.” It’s a commonly heard maxim. Over 60% of the respondents to the Parallel Audio Survey #1 checked “listening room acoustics” as the weakest link in the audio playback chain. It’s true, but it’s a little off the mark. The room is a passive element, not an active component in the playback chain. The room doesn’t make a sound. It can’t correct anything a system is doing wrong, nor can it improve the sound. It can, however, make it sound worse. Its acoustic behavior, its reflective and absorptive surfaces, and furnishings are worth paying some attention. The room contributes to the sound we hear, yet it isn’t entirely up to the room. A light analogy can help us understand why.

White light is defined as equal amounts of all visible frequencies. To properly judge the color of an object, we need full spectrum white light. If we have light that’s less than full spectrum or is deficient in some wavelengths, judging color is impaired, or impossible. For example, sodium vapor street lights radiate large amounts of yellow frequencies, almost no green or blue. Cars parked under sodium lights at night look strangely off-color compared to the color seen under midday sunlight.

Now, let’s do a thought experiment. We have a page of color swatches to choose from. Sitting in a closed windowless room, we turn on a light to view the swatches. Light bulb #1 produces full spectrum white light that radiates 360° around the room. We can judge the colors accurately because we have good full spectrum light. Change to light bulb #2. It radiates white light on one side, but as the we walk around it, the light becomes progressively redder. Now, judge the color swatches again. The light in front of the bulb, if we’re up very close, is good for judging, but on the backside it’s impossible. What about the overall color of the light scattered around the room? Despite the good light from one side, the red light from the back reflects off the walls, ceiling and floor, and mixes with the neutral white to cause a warm, reddish cast throughout the room. Our swatches, even on the white side, will be less accurately viewed.

We have something similar going on with loudspeakers. On-axis a speaker may be linear (neutral white), but as our ears move off-axis the sound becomes progressively lacking in higher frequencies (yellow moving to orange). At the back of the speaker the loss extends down into the midrange (leaving only red). So what? If we listen in an anechoic chamber, it wouldn’t matter. But our homes are not anechoic. When we listen to music, the speakers are illuminating the whole room with sound. The first arrival sound from on-axis is (or should be) good, linear, full spectrum sound. A few short milliseconds later, we hear the sound reflected from the room’s surfaces. This reflected sound includes all the acoustic energy radiated by the speaker, on- and off-axis, and becomes the sound of the room, or the reverberant soundfield. However, it’s not just the room we’re hearing. We’re hearing the speakers’ off-axis response. In most rooms, at typical listening distances, the reverberant soundfield makes up more than half of the sound we hear at listening position. Yes, more than half. The reverberant soundfield is a product of the speaker’s total 360° energy output. A loudspeaker’s total energy output is referred to as its power response. A nonlinear power response colors the sound of the music we hear. To exacerbate the situation, our rooms are not acoustically neutral, i.e., they don’t reflect the entire audible spectrum equally. Higher frequencies have greater losses; lower frequencies are less effected; for the lowest, there’s little loss (one of the causes of boomy bass). Add the colored room reflections to the colored power response of the speaker, and you get an earful of off-color sound.

Solution #1: build an anechoic chamber. It’s not too practical, takes a huge amount of space, and is enormously expensive. Very few audio companies, even the big ones, have one for those reasons. Far short of a full anechoic room is heavy room treatment. There are limitations to this. The frequencies that need the most control are the long wavelengths. Wavelengths a few meters and longer take heroic measures to absorb. Just look at the real world measurements of anything claiming to be a “bass trap.” They barely have any effect below 160 Hz. Even anechoic chambers are no longer anechoic below 80 Hz. That’s two-thirds of the bass range, more than that in practical terms since most musical content is above 35-40 Hz.

Solution #2: Nearfield listening. Set up your speakers with generous distances from the walls to delay reflections; and sit closer to the speakers than any of the ear-to-wall distances to keep the direct-to-reflected SPL (sound pressure level) ratio at least 1:1 or greater. Care should be taken with this technique. If you go too far, you’ll end up with a near headphone-like experience, spoiling stereo imaging and soundstage. This works best when you have a dedicated listening room. Still, even with nearfield listening, the better the power response, the better the reverberant soundfield.

Solution #3: A loudspeaker designed to be linear both on- and off-axis. It doesn’t matter if the sound level drops off-axis, only that it drops equally at all frequencies to maintain a linear power response. Another way of thinking about this is a speaker designed to behave well in normal listening rooms. It’s nearly impossible to do because of the wide range of wavelengths involved, approximately 20 millimeters to 20 meters. The behavior of short wavelengths is vastly different from longer waves. Short waves are easy to scatter or absorb with irregular or soft surfaces, but they also easily beam as a function of wavelength relative to radiating surface size. Long waves simply wrap around everything—the speaker itself, and even human-sized objects. Like being chest-high in the water at the beach, the waves go around you as if you weren’t there. There are numerous approaches to executing a linear power response : omnidirectional, dipole, controlled dispersion (via horns, waveguides, or large effective radiating surface). Each of these methods have their plusses and minuses. Dipole, horns, and other controlled dispersion are not necessarily designed for a linear power response, and may, in some cases, actually be worse by increasing the dispersion discrepancies and interference patterns.

There’s little discussion in the audio marketplace about power response. The majority of loudspeaker manufacturers won’t touch the subject, most haven’t considered it, many pretend it doesn’t exist. But our ears know better; they hear it everyday.

See the results and input your opinion on the Parallel Audio [survey].

Watch Floyd Toole’s video on [The Art & Science of Sound Reproduction].

For part 2 of the hearing series [Pro Gear].