Page 23 - Winter 2011

observation that the ear and brain can detect localization and timbre in a reverberant field more easily at the vocal formant frequencies than at the fundamental frequencies of most instruments. At low frequencies there are too few cycles in the brief time before reverberation overwhelms the direct sound. In addition the ability to separate sources into inde- pendent streams depends in part on the presence of multiple harmonics from the same source in each critical band of the basilar membrane—and this happens largely at higher fre- quencies. So if we can maximize the strength of the direct sound relative to the reflections and reverberation at high frequencies—while leaving the reflections strong at lower frequencies—we can achieve both good clarity and rich reverberation at the same time.
Second, the model predicts that clarity—source separa- tion—depends on the time delay between the onset of the direct sound and the cumulative sum of reflections in a 100ms window. The larger the time delay the greater will be the sum of nerve firings from the direct sound compared to the number from the reverberation. These predictions lead to a method of understanding the clarity of halls, and why their properties do not scale with size.
Listening in concert halls
It is widely believed that a shoebox shape is ideal for clas- sical music performance, regardless of the size of the hall and the type of music performed. Since the eye is tolerant of scale—a shoebox holds 2000 people as easily as it holds shoes—we assume the same holds for sound. But there are far more mediocre shoebox halls than great ones, and the small- er the hall, the poorer they are likely to be. In Leo Beranek’s surveys of musicians and conductors only three halls are rated “excellent.” Many people consider the Boston Symphony Hall (BSH), with 2625 seats, to be the best. But many of its close copies fall short. The odds of building an excellent new hall with a shoebox shape do not appear to be good—especially if the copy is smaller.
Other options exist. Currently “vineyard” halls are pop- ular. These halls are typically oval in plan, with no overhang- ing balconies. The audience surrounds the orchestra in ter- races (vineyards), the walls of which are intended to supply early reflections. The average listener is closer to the musi- cians than a typical shoebox of the same capacity, but many sit in poor seats behind the orchestra. Late reverberation in vineyard halls tends to be weak because the direct sound is either absorbed by the audience, or is directed down into the audience by panels on the ceiling (and is thus absorbed). There is little sound left over to create late reverberation. These halls lack the warmth and envelopment of BSH. A bet- ter option is exemplified by the Teatro Colón in Buenos Aires, which resembles a large semi-circular opera house. It is renowned as a concert hall where music is heard with extraordinary clarity and reverberation in a great majority of seats. But Beranek lists it in his books as an opera theater, and does not rank it as a concert hall. Jordan Hall at New England Conservatory, with 1013 seats, is a Mecca for chamber musi- cians and audiences from all over the world, and is also excel- lent for small orchestras.
Neither of these halls is a shoebox. Both are semi-circles with high balconies. They bring the average listener closer to the musicians than in a shoebox, and their high ceilings pro- vide the cubic volume and the reflecting surfaces needed for fine late reverberation. But citing the success of BSH the pub- lic, architects, and large donors usually demand a shoebox shape. BSH beats the shoebox odds by a rare combination of many elements, starting with size, shape, and surface. Changing any of these elements makes success unlikely.
Binaural recordings of live concerts in BSH reveal that the timbre and localization of each instrument is excellent everywhere on the floor—up to about row Z, just beyond the cross aisle, ~80 feet from the conductor. At the same time the reverberation is nearly always audible as the music is playing, lending a wonderful ambience to the sound. Surprisingly both clarity and reverberance are excellent—perhaps even better than on the floor—in the front row of the first balcony, 110 feet from the conductor. BSH succeeds in delivering both clarity and reverberation over a large majority of seats—a feat extremely rare in concert halls. How does it do it?
In most seats in BSH you can (with practice) hear all the notes in the music separately, and tell which instruments played them. You can also hear the reverberation of the hall as separate from the foreground notes. To hear all the notes separately your brain must be able to process the sound that comes directly from the instruments. But the direct sound is easily muddled by excess early reflections. Our neural model shows that the brain needs about a tenth of a second of direct sound to detect the pitch, timbre, direction, and distance of each player or section as separate from the others. If the number of nerve firings from the direct sound at the start of each note is greater than the number of nerve firings from the reflections for this critical tenth of a second, the brain can detect all the information we need. If the number from the reflections is greater in this tenth of a second the instruments blend together. The reverberation becomes the notes—audi- ble as chords, but not as separate sources. We still hear har- mony, but not the notes that create the harmony.
The kind of enveloping reverberation that gives a hall its richness is only audible if the direct sound is separately detected, and then only after the critical tenth of a second has elapsed. If the majority of the reverberant energy has decayed before this time the hall will be perceived as dry. In brief—we only hear all the notes if the number of nerve firings from the direct sound is greater than the number of nerve firings from all the reflections in the first tenth of a second, and we only hear the reverberation as separate if it is still strong enough to be heard after this period has elapsed.
The success of BSH is due to two factors: the audibility of the direct sound at frequencies above 1000Hz, and the rela- tively high strength of the late reverberation—sound that arrives at the listener more than 100ms after the direct sound. A major tenent of current acoustic design is that strong reflections from the side of the hall are essential for good sound. The experiments on which this tenent is based assumed that the direct sound was always audible, and only tested cases where the direct sound was nearly as strong as or stronger than the sum of all the reflections. Unfortunately in
Clarity, Cocktails, and Concerts 19