I Have Heard the Future...(DSP Room Acoustics Correction)
JGH began his talk by observing that high-end audio is no longer just a cottage industry. It has become a big business and a new "establishment," with money, influence, and its own trade association (the Academy for the Advancement of High End Audio). But Gordon, still a maverick after all these years, reminded the leading lights of this industry of an awkward truth: Although their stated goal of lifelike re-creation of concert sound has not been achieved, much of the industry has been focusing its attention on small differences while ignoring the large issues that continue to separate live from reproduced sound. Even non-audiophiles standing in an outside hallway can easily tell whether a sound is canned or live (footnote 1).
JA concluded the evening by emphasizing that, in order to make genuine progress, we must reinvent ourselves at regular intervals. High-end audio was very different ten years ago, he reminded us, and will be different a decade hence, in ways that we cannot foresee today.
Of course, the fireworks outside hadn't been ordered up for the occasion by our esteemed publisher; the City of Chicago was celebrating Memorial Day. Nevertheless, the rockets exploding colorfully in the sky were entirely appropriate: Stereophile's 30th Anniversary (and LA's tenth year as publisher) marked the maturation of high-end audio as an industry that now is not merely a fringe but a mainstream business.
The fireworks were appropriate for another reason. A new technology, which was unveiled at breakfast that day, represents nothing less than a true revolution in sound reproductionthe first real attempt to conquer the listening room. Kevin Voecks of Snell Acoustics demonstrated how digital signal processing (DSP) can be used to remove the sonic colorations caused by standing waves and boundary reflections. This breakthrough will allow listeners in many rooms to hear, for the first time, how good their loudspeakers and recordings really are.
Speaker designers have created transducers that turn electrical power into sound with splendid accuracy, both on- and off-axis. But speakers that sound great in one room may sound awful in another, or sound good at only one location in a roomif you're lucky enough to find it. We've all had the experience of buying speakers that sounded terrific in the store (or in the published review), but were disappointing at home. No matter how good a speaker is designed to be, in most rooms the sound waves arriving at your head are anything but accurate. If you measure your system's frequency response by placing a calibrated microphone in front of your nose, the variation in response from the highest peak to the deepest valley may exceed 20dB. These aberrations can't be corrected by analog equalization, because they are time-dependent.
The direct sound that arrives at your head in a straight line from the speaker is as accurate as the designer was able to make it. But if the driver is 3' above the floor and you are seated 8' away, a strong floor-bounce arrives just two milliseconds after the first-arrival wave. Due to this delay, at 280Hz the floor-bounce is out of phase relative to the direct sound when it gets to your ear, causing a broad measured cancellation notch around that frequency. Usually the floor-bounce is also severely irregular around the crossover frequency because it is produced by a wave-launch 35° to 40° below the speaker's axis, an angle at which the woofer and tweeter in many speakers go substantially out of phase.
During the next few milliseconds the arrival of the floor-bounce at your head is followed by other reflectionsfrom side walls, the ceiling, the wall behind the speaker, and the wall behind your chairall producing out-of-phase cancellation notches at other frequencies. If you were so unwise as to locate the speakers or your head at nearly equal distances from several of these reflecting surfaces, the cancellations will all occur at the same frequency, making the notch very deep indeed. This problem, named the Allison Effect after the New England speaker designer whose AES papers analyzed it, was discussed in these pages by the editor of the Boston Audio Society Speaker, David Moran (Vol.15 No.3, pp.1519).
During the next 100ms, sound waves bouncing back and forth between room boundaries will build up the patterns of reinforcements and cancellations that we call standing waves. The resulting peaks and valleys occur at frequencies that depend on the room dimensions. You can reduce their severity by experimenting with both speaker placement and listener location. (Low-cost computer programs such as Sitting Duck Software's The Listening Room help with speaker placement by displaying the frequency distribution of the peaks and valleys.) But as long as you have parallel walls, you will have standing-wave problems.
The basic low-frequency standing waves along the room's length and width have the greatest effect on the sound. If your room is 18' long (a typical dimension), the first fundamental resonance or "eigentone" occurs at 32Hz, producing a pressure maximum at each end of the room and a null in the middle. If you locate your speakers and yourself away from the walls (for the sake of better imaging), you probably won't experience the lowest bass that the speakers can produce. If the width of your room is around 12' (the most common dimension in a house or apartment), the fundamental standing wave on that axis occurs at about 50Hz. It's likely that your speakers are symmetrically placed in the room and you probably sit on the stereo axis, meaning that your head is midway between the sidewalls. In that case you will experience a sound-pressure minimum at 50Hz and a sound-pressure peak near 100Hz. This combination of weak low-bass and turgid midbass is the single most common fault in domestic hi-fi. Deep bass may not be important to you, but dense midbass will color nearly every recording you play.
Footnote 1: JGH's talk was published as "As We See It" in our September 1992 issue, Vol.15 No.9.John Atkinson