As We See It

If there's a technical thing nearly every audiophile knows, it's that low frequencies (LF) aren't directional. We tend to treat LF as mono—think sub/sat systems in stereo, "LFE" in home-theater. But if the conclusions of a study by Thomas Lund, a researcher at Genelec OY (footnote 1), imply what they seem to, then that conviction is quite wrong.

In addition to resetting that bit of conventional wisdom, Lund's study may have uncovered an important contributor to the feeling of envelopment so many audiophiles crave, itself a profound source of pleasure beyond the music itself. As I have often said, and probably written once or twice, listening to a good hi-fi system is like getting a massage.

I wrote about some of Lund's previous work in this same space five-plus years ago; he's the person responsible for the phrase "slow listening." His new work focuses on the experience he calls Auditory Envelopment (AE). AE is not exactly the same as LEV—listener envelopment—a more familiar concept from the scientific hi-fi literature. Lund's phrase refers to our response to "low-frequency interaural change," which, he has found, increases the sense of envelopment in music—which we feel "when sharing a large acoustical space with an orchestra or a choir."

Lund and colleagues recruited "naïve" participants—people who had not previously worked in music or audio—and did several related experiments. First, they played pink noise with a sharp low-pass cutoff of 150Hz. The left and right channels were, alternately, correlated and uncorrelated (footnote 2). They asked the naïve subjects, can you hear a difference? All of them could.

Next, they asked the participants to describe what they heard. In broad terms, correlated LF sound was reported to be unpleasant, claustrophobic, associated with a small, restricted space. In contrast, uncorrelated LF sound "was found to be friendly and associated with a large or unlimited space." Words used to describe correlated sound revolved around 'lille,' 'lite,' 'spærret inde,' 'låst inne,' 'uhyggelig,' 'skummelt,' 'mørk.'" Those terms are Danish for small, restricted, locked up, spooky, dark.

Words used to describe uncorrelated LF sound included "stor," "åben," "åpen," "fri," "behagelig," "lys," which translate as big, open, free, pleasant, light. Correlated LF noise = bad. Uncorrelated LF noise = good.

Finally, they progressively lowered the upper cutoff frequency of the pink noise until none of the participants could hear the difference between the correlated and uncorrelated LF noise. At 54Hz, several participants could still hear the difference. At 30Hz, none of them could, suggesting that AE isn't a bottom-octave phenomenon.

Lund considers AE parallel to something called affective touch. If I understand this, affective touch means little more than that being touched can feel good, even though touch is routinely used analytically—to read braille, for instance, or to play the piano or violin. In one of the best (if ultimately failed) efforts I've seen to use scientific language to render sexy unsexy, Lund writes, "Skin responds to stroking of just the right force, speed, and temperature, triggering a reward reaction from the striatal cluster in the brain of the receiver. This causality is now widely accepted and regarded a potentially socializing mechanism of friendly touch." (He obviously means more than just sex here, but hey.) Under ideal conditions, then, listening to hi-fi is akin to sex.

Lund's article is less technical than many technical articles; still, it took me a while to get the gist. To make sure I had it right, I sent him an email with my summary.

Me: We minimize the importance of what I'll broadly call "stereo effects" at bass frequencies. That seems like a logical conclusion from the facts that bass in rooms is wavelike, not raylike, and we can't hear direction at those frequencies anyway. But your article suggests that obscuring L–R differences at low frequencies robs music of its sense of envelopment. One obvious implication is that the whole sub/sat approach is fundamentally wrong, as is LFE [the mono low-frequency effects channel] for music, using a single subwoofer, or randomly placed subs in a room.

Lund: Exactly.

So if you've got a sub/sat system at home, go buy a matching second sub to form a proper L–R pair, place your subs near (preferably just to the outside of) your main speakers, keep the crossover point low, and tap the sound from the main speakers for the two subs.

Some details: Lund's results would seem to suggest that a reverberant listening room is a good thing, since uncorrelated low-frequency energy increases the sense of envelopment. In fact, the opposite is true: The listening spaces best able to reproduce the AE inherent in a recording are those that add the least uncorrelated LF noise; you need a dry room to create a wet experience. This assumes that the information needed to produce AE is captured by the recording. Often it isn't. On recordings where it isn't, a few extra reflections in the playback space may help mitigate that deficiency.

"A good playback room mediates the recording without overly harming potential AE with obnoxious room modes," Lund wrote in our email exchange.

Other potential AE assassins include lossy compression and overly aggressive DSP in the bass. "Upstream, lossy data reduction or bass mismanagement can already mean 'game over.' As a sidenote, more LF L–R difference is possible in lossless digital than with vinyl, though pickup rumble actually isn't in phase." It's possible then that turntable rumble could increase the sense of envelopment. "That effect might be interesting to study also."

Lund told me that while "3D music"—Atmos, etc.—has more potential for this kind of envelopment, it's also more vulnerable to lossy compression. "3D distribution systems are generally more poisonous to AE than stereo codecs—down to [a bitrate of] 256kbps at least."

Broadly, unsubtle attempts to achieve flatten low-frequency response work against AE—which is not to say that room resonances should be left alone. Of all the interesting things Lund said in our email exchange, I like this quote best: "Room modes dampen AE, often taking out entire octaves of potential joy."

---

Footnote 1: The paper and the associated presentation can be downloaded from bit.ly/LundEnvelopment.

Footnote 2: "Correlated" means that the L and R channels are "in phase"; with noise, it means that the left and right sound is exactly the same. "Uncorrelated" means that while the bandwidth and levels of noise are exactly the same, the noise sent to the left channel is independent of—randomly related to—the noise in the right.