(2) Designed by Walker and Weeks Architects, Severance Hall has an Indiana limestone superstructure cladding and an Ohio sandstone base. Its massive construction provides a prohibitive and effective barrier to exterior street and mechanical equipment noise. The hall form was based upon the goal of providing the best acoustics for an audience of approximately 2000 people. Construction on the hall began on November 4, 1929, days after the Wall Street Crash. The opening concert occurred on February 5, 1931.
(3) The hall combines eclectic elements of classical, Art Deco, and Egyptian revival design. Marble and bronze accentuate an elaborate foyer while the subdued auditorium is covered by plaster and draperies. Several auxiliary spaces recall an elegance and style reserved for a select few, but now modified to serve a broader audience and achieve greater economic benefit. An example is an original vehicular drive-through entry for up to fifteen limousines at one time, now used as a restaurant.
(4) Walker and Weeks engaged Professor Dayton C. Miller, a close personal friend of Professor Wallace C. Sabine, as the acoustical consultant. Miller espoused the typical feeling of acousticians around the year 1930, that is, that existing auditorium reverberation times were too high. The result was that the original Severance Hall was designed to have a small volume (550,000 cubic feet) to surface ratio with elaborate seats, carpets, drapes and furnishings, producing a short reverberation time. Measurements made by A. H. Benade and R. S. Shankland in 1953 indicated a reverberation time of 1.5 seconds for the original auditorium at frequencies of 250 and 500 cycles per second (cps). Today, this value is generally perceived to be a compromise between the optimum situations for music and for speech.
(5) Initial reactions to the acoustics were quite favorable. But as time progressed and the quality and reputation of the Cleveland Orchestra increased, the professional musicians became increasingly aware of the low reverberation times, which affected particularly the higher frequencies. Dr. George Szell became musical director of the Cleveland Orchestra in 1946 and his awareness of the quality of comparable concert halls in both the U.S. and Europe, together with his expansion of the traveling tours of the orchestra, made comparisons inevitable. In 1953, R. S. Shankland and A. H. Benade tested the reverberation times, analyzed the situation, and made a number of recommendations. It was not until 1958, however, that a significant number of the changes were made.
(7) The changes eventually made in 1958 included a new wooden shell stage enclosure with surface walls built as a series of convex five-foot, six-inch-radius plywood curves and a second layer of flat plywood panels behind the curved front surfaces. To cut resonance at low frequencies, the space between the curved fronts and the flat backs was filled with sand to a height of nine feet. The proscenium arch was removed. The remodeled stage balanced an increased reflection to the audience and a thorough mixing and blending of sound on stage. Also, the carpeting in the auditorium was removed and replaced with vinyl tile.
(8) During the 1958-59 concert season, Shankland, Benade and Helmut Ormestad measured the new reverberation times. Sound sources included a revolver, the full orchestra in an empty hall, and the full orchestra during performances. Measurements were taken on the main floor, in a box seat and in the balcony. "The results show a considerable increase in the reverberation times at all frequencies as compared to the conditions existing prior to 1958. The (new) reverberation time with a full audience is about 1.7 seconds in the mid-frequencies and rises steadily to well above two seconds at lower frequencies"Ref.2. Due to the heavy cushioned audience seats, the 100-person orchestra makes a greater impact in reducing reverberation time than does the 2000- person audience.
(10) Shankland's reports Ref. 2, 3 are unclear as to which audience positions were used to do the measurements. He mentions three positions, but reports only one set of reverberation times. Perhaps, the values represent an average condition, an approximation for one position.
(11) The 1958 tests were made at frequencies from 150 to 6400 cps. The healthy human ear has the capability of hearing sounds from 20 to 20,000 cps. Orchestral sounds from 50 to 11,000 cps are common.
(12) Electronic sound sources covering the full sound spectrum with controlled amplitudes are now available in place of revolver noise and a particular performance of the full orchestra. The earlier measurements were executed via audio tapes transported from the auditorium to a testing laboratory. The tapes were generally inadequate in duplicating the 30- to 60-decibel drop necessary for meaningful documentation of reverberation times.
(13) As new electronic technology has emerged and provided the capability to document large amounts of data, the accuracy and precision of the traditional reverberation time formulas (specifically, the Sabine and Eyring formulas) have been called into question. Directionality of the sound source and a variety of surface positions with respect to the sound source with varying absorption characteristics are not recognized by the formulas. Also, laboratory information regarding absorption characteristics of surfaces at different frequencies is woefully inadequate. Actual measurement techniques can provide not only more data but more precise data.
(14) Madaras recognized not only the improved technology and methodology available thirty years later, but also that the air conditioning system control was much more stable today and that the seat surfaces had a considerable decrease in absorption capability since installation in 1948. An increased reverberation time would be expected.
(15) With the cooperation and assistance of representatives from Bruel and Kjaer instruments, Inc., acoustical measurements were taken February 4, 1991, in Severance Hall. The hall was empty with the exception of staff and five technical people conducting the tests. Lights and HVAC were operating. The measured ambient air conditions were 75 degrees Fahrenheit and 40% Relative Humidity. Measuring equipment included:
On center stage: B&K 4224 electronic sound source (a full spectrum white noise) JBL amplifier, property of Severance Hall Two-way speaker box, property of Severance Hall At side of stage, monitored by technical personnel: B&K 2133 dual channel real time frequency analyzer In test positions in auditorium seating: B&K 4230 sound level calibrator B&K condenser microphone on a tripod.
(16) Measurements were taken at twenty-four seat positions, including eleven on the main floor, four in the box seats, five in the dress circle, and four in the upper balcony. The generated sound source included frequencies from 12.5 cps to 20,000 cps sustained for 1.0 second at which point the reverberant sound reached maximum sound pressure. At this point, sound was abruptly stopped.
(17) Measurement of sound pressure level was recorded every thirty milliseconds, beginning 360 milliseconds before the sound was stopped and continuing for a duration of 3.0 seconds. It was judged that after 3.0 seconds, sound would have decreased to the background noise level. Thus, data were recorded 100 times for each source signal and repeated ten times for a total of 1000 measurements per frequency. Sound energy decay was recorded for each one-third frequency spectrum. This process was repeated ten times at each measuring position and the results averaged to a smooth decay curve.
(18) The first twelve records occur during the signal and at the thirteenth record, the source is stopped and amplitude begins to deteriorate until background noise level is reached. This occurs for each of the thirty-three third-octave band frequencies and at each of the twenty-four measuring positions. Over 792,000 amplitude readings were taken over a period of three hours.
(19) The data were processed using a computer program which produced three-dimensional graphs for the entire frequency range (12.5 to 20,000 cps), time (0 to 3.0 seconds), and amplitude (110 to 0 dB). Two-dimensional relationships of amplitude versus frequency revealed sound distribution patterns for each position. At 3.0 seconds this relationship revealed the characteristics of the background noise. The two-dimensional relationship between amplitude and time describes reverberation times. (see accompanying figue.)
(20) Reverberation time is defined as the amount of time taken for a sound to decrease in amplitude 60 dB. With background noise levels in the range of 25 to 35 dB, a sound source would be required to produce an amplitude over 120 dB (above the threshold of human pain) to allow for deterioration over distance and provision of a measured sound of at least 60 dB above the background noise. Therefore, amplitude drops of 30 dB, 20 dB and 10 dB are often used and the data extrapolated to define a 60 dB drop. If the sound deteriorates in a linear fashion, this technique works well. If, however, there is a non-linear deterioration, questions arise as to the validity of such extrapolations. Many acousticians and musicians would agree that 10 dB and 20 dB drops are really more significant than a 60 dB drop because they express the quality of the earliest sounds that reach the ear of the listener.
(22) What we have gained is an increased understanding of how our individual experiences are achieved and how we can seek to connect our likes, desires, and needs to the analytical information. The true evaluation of the space involves the totality of all data, generating a new combination each moment.
(23) Analysis of the information and graphs achieved in this study led to the following conclusions with regard to relationships between audience seat position and reverberation time. The main floor can be divided acoustically into three zones. The center section has a reverberation time, 60 dB drop, of 1.8 seconds at mid-frequency. This low value is probably due to the lack of side-wall reflections and a ceiling form that directs sound toward the upper balconies. Primary reflected sounds appear to come from the stage.
(24) The second zone on the main floor is the seats between the center zone and the edge of the box seats of the lower balcony. Reverberation times of 1.7 to 1.8 seconds at mid-frequency occur here. Early side-wall reflections are received in this zone from a promenade which surrounds the main floor seating. The third zone comprises those seats on the periphery, under the lower balcony. Reverberation times of 1.8 to 1.9 seconds were recorded here.
(25) The promenade causes some peculiar standing waves as the sound bounces between floor and ceiling, a distance of eleven feet, a wavelength due to a 100 cps sound.
(26) The dress circle or lower balcony had a reverberation time of 1.9 seconds at mid-frequency, the most desirable conditions in the auditorium. The form of the ceiling and side walls contribute to the reverberation time.
(27) The upper balcony, with a lower ceiling, has a reverberation time of 1.7 seconds at mid-frequency.
(28) The criteria for reverberation times vary by performance and performer. What becomes clear from this analysis is that the design and operation of a great acoustical environment is still an art and is to be judged by the ear of the beholder. Severance Hall is one of those special places that elicits this kind of investigation and interest. It is far superior to any nearby auditorium and provides a special kind of experience. Its history, its occupants and its architecture combine to make it what it is. Hopefully, this study and similar studies will allow us to not only understand its acoustical character but maintain a desire for excellence and quality in continued artistic performance and efforts to improve.
Ref. 1: Madaras, Gary Scott. "Ideal Reverberation Time Curves for Classical and Romantic Symphonies: An Acoustical Analysis of Severance Hall, Cleveland" (M. Arch. thesis, Kent State University, 1991).
Ref. 2: Helmut J. Ormestad, Robert S. Shankland, and Arthur H. Benade, "Reverberation Time Characteristics of Severance Hall," Journal of the Acoustical Society of America 32, no. 3 (March 1960), 371-375.
Ref. 3: Robert S. Shankland and Edward A. Flynn, "Acoustics of Severance Hall," Journal of the Acoustical Society of America 31, no. 7 (July 1959), pp. 866-871.
Accompanying graphic reproduced from Madaras (1991), Fig. 14.
Material appearing in Architronic may be distributed freely by electronic or any other means, providing that any such distribution is without charge (unless for purposes of cost recovery by interlibrary loan services) and that Architronic is acknowledged as the source. However, no article may be reprinted in any publication without the explicit written permission of the author(s). This statement must accompany all distributions of Architronic, whether complete or partial.