Page 12 - Winter 2011

neatly into typical parametric categories.
The imbalance among different instrumental sections of
an orchestra is one example of the shortcomings of our exist- ing parameters. There is no widely accepted parameter to describe orchestral balance, probably because it is influenced by such a large number of factors including accuracy of acoustics at the conductor’s position, musician placement, riser configurations, instrument directivity, musician playing technique, repertoire, and specific sound reflection paths in the room. In a situation as complex and subjective as this, it seems unlikely that any single parameter for orchestral bal- ance could emerge. Yet the issues remain critically important. We suggest that significant new progress in concert hall design will require delving into complex and multidiscipli- nary issues such as these.
When has ‘average’ ever described something of extraordinary beauty?
Another problematic aspect of our room acoustics parameters is that they tend to be excessively averaged over space, time, and frequency. While making measurements at a limited number of seating positions is a practical reality in the consulting world, it is too easy to assume that the room is “well-characterized” even if the number of measurement positions greatly exceeds those outlined in the ISO 3382 stan- dard. Understanding the differences among various seating locations may be even more important than their averaged values. The Swiss Alps would be of little interest if we observed only average values—a high plateau covered with rocks, pine needles, and alpine flowers under a thin layer of snow and ice.
Parameters that depend strongly on direct sound and early reflection structure can be especially misleading if aver- aged over seating locations, since the most interesting and significant qualities of these parameters are often how they vary within the room. The variation of Early Decay Time (EDT) in a hall with deep under-balcony areas is a good
2
example of this. A single room-averaged EDT value, as typi-
cally reported, would entirely miss the relevant, and perhaps obvious, point that a listener will feel disconnected from a full sense of enveloping reverberation if seated under a deep overhang.
Unoccupied vs. occupied data
Acoustics characteristics in halls change when musicians and audience are present. Yet the vast majority of room acoustics data available for performance spaces is based on
3
unoccupied measurements. While some data for occupied
halls is available, not enough is known about the influence of an audience and/or orchestra on the acoustics parameters, especially on those other than reverberation time. All param- eters that are influenced by the strength of direct sound are rendered inaccurate if they do not account for the fact that audience heads obscure listeners’ access to that sound.
Modern multi-channel measurement systems make it feasible, however, to acquire impulse response data efficient- ly and simultaneously at multiple positions in an occupied concert hall. With careful planning, the audience and musi-
cians need to remain quiet only 20 to 30 seconds while the test signals are played and recorded in many positions in a hall. In our experience, audiences find this measurement process to be entertaining, especially if there is a very brief explanation of their role in changing the acoustics and the reciprocity of noise transfer from the audience to the stage. Crinkling a candy wrapper in front of a very quiet audience dramatically demonstrates that if gentle pianissimo sounds of the orchestra are heard everywhere in the hall, the quiet sounds of opening a cellophane-wrapped candy are also per- fectly audible everywhere in the hall. Such brief moments provide unique opportunities for outreach and education to the general public. When concertgoers and musicians are engaged in the measurement process, we acquire much need- ed acoustical data for occupied halls. Moreover, audiences become engaged directly in meaningful research aimed toward making real improvements for both the audience and the musicians.
A different measurement approach
We believe that a useful approach for concert hall acoustics investigations is to focus on understanding the spe- cific phenomena that determine the character of a hall, rather than focusing on a hall’s statistical characterization. Within this framework, we first listen to music carefully in the room—ideally during a rehearsal—and note the good, not- so-good, and bad acoustical/musical qualities throughout the space. After listening to music, we explore the unoccupied room with a steady-state full spectrum sound source placed on the stage. This greatly facilitates listening for timbral bal- ance, evaluating evenness of sound distribution, and identi- fying particular (sometimes anomalous) reflection paths. Since a steady-state sound source does not easily reveal arrival time of reflections, listening also to a loud, sharp- clicking metronome is very useful for understanding time domain characteristics. Only after we develop a full grasp of the various subjective qualities and variances of a space do we pursue instrumented measurements. Those measurements seek first to unravel any mysterious observations before doc- umenting what may be a ‘typical’ condition. To the extent possible, such observations and measurements are ideally made in both the unoccupied and occupied hall.
The goal through all this is to understand the space as a whole as well as the assemblage of its parts. When we identi- fy particular seats that are notably good, bad, or just ‘unusu- al’ acoustically, we measure room impulse responses at those locations often with a highly directional listening and record- ing device to discover and help extract the most valuable information in the data—perhaps a late-arriving reflection, a series of reflections with limited frequency response, or an area of grossly uneven sound distribution.
A curious example
An anecdote from recent experience in a well-known opera house may serve as a useful example of this phenome- nological, rather than statistical, approach to acoustics. In this case, we had been made aware of concerns about the sound of the orchestra in the pit being overly loud at certain
8 Acoustics Today, January 2011