3.1. Measuring Method
The measurements were carried out when the hall was unoccupied. A Bruel & Kjaer sound level meter and analyzer type 2250, DIRAC room acoustics software type 7841 [20
], omnipower omnidirectional sound source, and a power amplifier with a receiver and transmitter were used. The position of the source and twenty measurement points are shown in Figure 3
. The layout of the measurement system is given in Figure 4
DIRAC is capable of producing internal noise of various types and analyzing the impulse responses of the room by the use of a PC, a sound device, and microphones at heights of 1 m, 3.5 m, 5.5 m, and 7.5 m, as shown in Figure 3
. In this study, Maximum-Length Sequence (MLS) was used as an alternative to the conventional white noise. The noise was produced in the room by the omnidirectional loud speaker through a power amplifier at a hieght of 1.5 m. DIRAC then calculated the frequency spectrum and many other acoustical parameters, such as reverberation time, early decay time (EDT), clarity (C80), definition (D50), and speech intelligibility. Acoustical parameters were determined according to the ISO 3382-1 and BS EN 60268-16 standards [21
There are a number of objective and subjective parameters to judge the acoustic performance of a hall. The research has shown that some of these parameters have lost their importance. Generally, the most important objective room acoustic parameter is the reverberation time (RT). This is the time needed to achieve a 60 dB level decay after sound emission is interrupted. It gives fullness and a pleasant tone to music but must not be so great as to destroy voice intelligibility. While it is a very important parameter of room acoustics, it is not the only one. The relationship between a sound falling and the time needed from when the sound source is turned off is not linear. Thus, it is necessary to find the slope of the fall such as T10, T15, T20, and T30 and convert it to a reverberation time of 60 dB. T10 is the reverberation time found by the slope of a fall between −5 to −15 dB, T15 is the reverberation time found by the slope of a fall between −5 to −20 dB and so on. Besides finding the reverberation time by T10, T15, T20, and T30, it is also necessary to find the best reverberation time, designated as the RT.
The early decay time (EDT) is the initial portion of the sound decay curve of the reverberation time. Kuttruff [23
] has shown that EDT is responsible for our subjective impression of reverberation. It affects the auditorium’s support of voices and adds definition to the higher tones of music. Similar to RT, if it is too high, it reduces intelligibility. RT is determined from the first 10 dB range of the decay curve. From the corresponding slope, the EDT is calculated as the time to reach −60 dB. EDT is also the parameter that relates best to speech intelligibility reduction.
The clarity (C80) is the logarithmic early-to-late arriving sound energy ratio, where “early” means “during the first 80 ms” and “late” means “after the first 80 ms”. It characterizes the separation in time of sounds generated by individual instruments or groups of instruments. It should be appropriate to enable musical detail to be appreciated.
The definition (D50) is used for speech signals and it is the early-to-total arriving sound energy ratio, where “early” means “during the first 50 ms”. The Definition is also called “Deutlichkeit”. There is a good relation between definition and speech intelligibility and definition is expressed as a percentage.
In the design of music auditoria, there are other parameters that have to be considered, including: Intimacy, Spaciousness, Diffusion, Warmth, Brilliance, Reflecting Surfaces, Spatial Impression (SI), and Stage Support Factor (ST1) [24
]. Intimacy exists if the music in an auditorium is heard the same as it is in small halls and if listeners feel as if they are in contact with the performers. In order to achieve this, the arrival time of the first reflection after the direct sound should be less than 25 ms. Spaciousness is achieved if the sounds at listeners’ seats include some early reflections from near lateral directions. Its measure, the inter-aural cross correlation coefficient (IACC), should exceed 0.6 [25
]. This parameter, however, was not measured in this hall. Diffusion in an auditorium is provided by large and small irregularities on the walls, balcony parapets, and the ceiling to give the sound a rich patina. Warmth is the presence of strong low frequency or bass sounds in a hall. Brilliance is the presence of strong higher frequency or treble components. Spatial Impression (SI) is the feeling of being enveloped within the music. It is determined by differences in the signal received by each ear. Barron and Marshall [26
] proposed an equation, for the degree of SI, based on Lateral Energy Fraction (Lf
) as: SI = 14.5 (Lf
Stage Support Factor (ST1) is a measure of the strength of the orchestral sound returned by nearby reflecting surfaces to the ears of each player in the orchestra pit. Beranek [10
] argues that stage support factor (ST1) is very important for musicians, but since this is a multi-purpose hall, nothing could be done to provide stationary reflectors around the stage. In some good auditoriums, it was reported that the ST1 is lower than −15 dB.
Reverberation time for speech auditoria of around 38,000 m3
should be 1.35 s, and for the music auditoria of the same size it should be 2.1 s for frequencies of 500 Hz and higher. For 125 Hz, these values should be increased by a factor of 1.5 and by 1.15 for 250 Hz [27
]. Since it is not practical to fit variable sound reflectors and absorbers to this hall, the authors adopted the optimum reverberation for the mid frequencies in the range of 1.6 to 1.8 s [28
], 1.4 to 1.9 s for the early decay time, and a bass ratio greater than 1.0. [19
]. Clarity (C80) should be between −2 to +2 dB and values of definition (D50) in excess of 0.50 are considered satisfactory for speech intelligibility [29
The other parameters for speech auditoria are the Speech Transmission Index (STI), Alcons, Room Acoustics Speech Transmission Index (RASTI), The modulation transfer function (MTF) describes to what extent modulation m is transferred from source to receiver, Articulation Index, Percentage Articulation Loss of Consonants, Signal-to-Noise Ratio (SNR), and Direct-to-Reverberant Ratio. In this study, only the STI and RASTI were considered because they are the newest parameters and can be derived by the use of DIRAC [20
]. Speech Transmission Index (STI) is a measure of the speech intelligibility in a room and is judged as unintelligible between 0.0 to 0.3, poor between 0.3 to 0.45, fair between 0.45 to 0.6, good between 0.6 to 0.75, and excellent when > 0.75. It can be measured separately for male and female voices. Whenever there is too much background noise, the Useful-to-Detrimental Index U50, suggested by Bradley [30
], may be an alternative to STI. Room Acoustics Speech Transmission Index (RASTI) is another speech intelligibility index. It is the simplified version of STI. In order to make correct RASTI measurements, the overall system frequency response must be uniform from 125 Hz through the 8 kHz octave band, the background noise must be smooth in time and frequency, the space must be substantially free of discrete echoes, and the reverberation time must not be too frequency-dependent [22
]. The average RASTI for all the points of the hall was 0.3435. Measurements of these parameters for a number of points are given in Table 1
. The results show that except for points near the speaker, speech intelligibility is very poor.
The STIPA is a simplified version of the STI (speech transmission index), intended to emulate STI under conditions typical for public address systems. Since STI measurements were made with DIRAC [20
], it was found that there was not much difference between STI and STIPA.
Impulse to noise ratio (INR) is defined as the logarithmic ratio of the maximum impulse response level and the noise level and reflects the decay range. This parameter is not a measure of room acoustics, however, it indicates whether the related measurements are correct. With good measurements, most practical INR values range from 35 to 45 dB [21
]. The INR in Figure 5
shows that the measurements are reliable.
Signal-to-noise ratio (SNR) is the signal of the speaker’s voice in decibels, minus the background noise. If the SNR is large, speech intelligibility is greater. If the SNR is less than 10 dB, speech intelligibility is degraded [31
]. Figure 5
shows that the SNR of the unoccupied hall for low and very high frequencies was lower than 10 dB. When the hall is occupied, it is expected to be lower than 10 dB for most of the frequencies.
The Bass Ratio (BR) or Bass Strength is an objective measure of warmth. It shows how low frequency (bass) sounds are perceived in a concert hall. This is simply the ratio of the low frequency to mid frequency reverberation times. Thus it can be expressed as: RT125Hz+250Hz/RT500Hz+1000Hz.
Most practical BR values range from 0.1 to 10. Bradley, Soulodre, and Norcross [32
] give a very good interpretation of BR and conclude that although the perceived strength of bass sounds is not influenced by the low frequency reverberation time, the levels of both the early- and late-arriving bass sounds influence the perceived strength of the bass sounds. The average Bass Ratio of all the points in the hall was measured as 0.94. In both speech and music auditoriums, there should not be any echo or flutter echo, and the sound energy must be diffused homogenously all over the auditorium. The measurements were made at the hall when there were not any spectators. The results of the measurements are given in Figure 6
In an auditorium, background noise is generated from the air conditioning, ventilation systems, machines, audience, traffic outside the building, and environmental causes like rain and thunder. It was important to measure the background noise level (BNL) for some of the parameters. If the BNL is high, the speaker may increase his/her voice to overcome the noise, although this may also cause mental pressure for the speaker. The background noise level (BNL) was measured as an Equivalent Continuous A-Weighted Sound Level when the hall was unoccupied but with the air conditioning system on; it was measured as Leq = 42.4 dB (A). It is very difficult to measure the background noise with the presence of spectators because it varies according to their level of attention. In another measurement, the hall was full with the spectators, the air conditioning system was on, and the spectators were asked for one minute of silence. The measured Leq was 50.0 dB (A) during that time.
Discussion of the Measurement Results
Since this is a multipurpose hall, the parameters related to both music and speech were considered. It can be seen from Figure 6
that for frequencies between 125 to 2000 Hz, the reverberation time was longer than 4 s. This is a very long RT for this auditorium. Parallel to the RT, the EDT was also very long. However, the difference between the RT and EDT was within the limits of 10% [10
]. In fact, Beranek [10
] has shown that EDT measured in an unoccupied hall is an even better indication of acoustic quality than RT. All these measurements were made when the hall was unoccupied as it is very difficult to measure acoustical parameters when the hall is occupied. The RT for the occupied hall was calculated according to the Hidaka et al. [33
] method for the known upholstery of the seats, and the results are listed in Table 2
. The RT was very long and unacceptable for both occupied and unoccupied conditions.
Although clarity (C80) indicates whether the musical details and the instruments in an orchestra will be heard pleasantly, it is also an indication of speech intelligibility. The average C80 in the auditorium was within practical limits and very close to the optimum range of −2 to +2 for multipurpose auditoriums. When the hall was empty, the C80 for frequencies between 125 to 2000 Hz was approximately −3 dB; this is lower than the acceptable level. Hidaka et al. [33
] have shown that C80 does not change significantly when the audience enters a hall.
Definition (D50) is more related to speech intelligibility than (C80). For all the frequencies, definition (D50) was lower than 0.5 s, thereby indicating inferior speech intelligibility. Another parameter related to speech intelligibility is the Signal-to-Noise Ratio (SNR). When the auditorium was unoccupied, the SNRs for the mid frequencies were more than 10 dB, which is a satisfactory level. SNRs were lower than 10 dB for low and high frequencies, especially for 31.5, 63, 4000, and 8000 Hz and higher for the mid frequencies. However, when the auditorium was full with an audience, the background noise increased and SNR fell to an unacceptable level, thereby degrading the speech intelligibility.
The measurements of STI and RASTI as seen in Table 1
show that, except for points very close to the stage, these speech intelligibility parameters were much degraded.
Normally, side reflectors are used in music auditoria to strengthen the singers’ voices. Studies by Barron and Marshall [13
] showed that “lateral reflections give a listener the subjective impression of being enveloped by the sound and they increased the spatial impression”. Two recent studies on the subject indicate that lateral reflections are beneficial due to their increasing effect on binaural loudness [36
]. In speech auditoriums, it is necessary to provide equal loudness all over the hall with side and top reflectors. It is not possible to use special reflective surfaces in this hall, because all the present reflecting surfaces cause an echo, and if side reflectors are used, they will cover the spectator tribunes during the sporting activities.
All these analyses explain the poor quality of acoustics for both music and speech in the hall, which is not helped by the inflexible seating arrangements. Our proposals seek to improve the acoustics within the context of the existing seating layout illustrated in Figure 1
The computer simulation of the hall was performed using the ODEON 14 software package [38
]. Twenty receivers were used in total with the sound source at the stage as shown in Figure 3
. ODEON 14 is based on sophisticated methods of physical analysis known as the modulation transfer function, image-source method, and ray tracing. With simulation models, it is possible to see the effects of different geometries and the effects of different surface materials on room acoustics. In this study, the authors checked the latter. The first simulation (Simulation No. 1) was conducted for an unoccupied condition. Sound absorption coefficients of all the surfaces used in the simulations are given in Table 3
and the simulation parameters are given in Table 4
. This simulation was done in order to find the similarities between the measured and simulated values, for which the just noticeable difference (JND) was used as the unit of measurement. The measured and simulation values of a number of parameters show agreement, especially between 500 to 2000 Hz frequencies. The small differences observed are due to the assumptions of the sound absorption coefficients of the surfaces in the simulations. Outside this range, there is a significant difference between the measured and simulated values. The next simulation (Simulation No. 2) was made for the occupied hall. The averages of all the twenty points when the hall was occupied and without any treatment are given in Table 5
. In the simulations, an attempt was made to use a curtain with a surface density of 0.26 kg/m2
all over the seating area instead of with an audience, as proposed by Hidaka et al. [33
]. In this simulation, the results were not more satisfactory and so the idea was not used.
In Simulation No. 3, the ceiling and walls were covered with acoustic plates. It was decided to glue grey colored, 5 cm thick fireproof pyramidal melamine plates on the ceiling at the bottom surfaces of the aluminum sandwich panels. Melamine plates are lighter and more durable then polyurethane plates of the same type. On the walls, cream colored melamine plates with fireproof textile on the front surface and an air cavity at the back were added. So, the colors of the ceiling and walls were not changed and the red aluminum space-frame elements were kept visible. These elements would, at the same time, contribute to the diffusion of sound during concerts. The results are given in Table 5
. This treatment provided too much absorption, and the reverberation time was reduced too drastically. Consequently, it was decided to reduce the sound absorbers in the fourth simulation (Simulation No. 4). In the fourth simulation the back wall and all the walls of the side tribunes were kept as they were without any acoustic treatment. Thus, the narrow columns, beams, and the long narrow horizontal walls in this area will increase the diffusion of sound, which is a favorable condition during music activities. The results are given in Table 5
. This provided better RT and EDT values. In the hall, some curtains were used to cover the glass surfaces and it is also recommended that these curtains be open during the music activities.
Discussion of the Simulation Results
The discussion below is for the fourth and final simulation. The average reverberation time of the occupied hall for the mid frequencies with the proposed measures was 1.6 s. Thus, the required criteria were met in terms of RT for both speech and music. The average EDT for the mid frequencies was 1.22 s, which is below but very close to the 1.4 to 1.9 s range. This is due to the absence of side reflectors in this hall. The requirement that EDT should not be more than 10% lower than RT could also not be met and was 20% lower.
C80, which is an important parameter for both music and speech, remained between −1, 0, and 10 dB for all frequencies. For the mid frequencies, it was approximately 4.0. While these are practical values, it is impossible to provide the ideal values of −2 to +2 dB. Another speech intelligibility parameter D50 reached the criterion of being above 0.50 except for very low frequencies. STI results after the proposed measures would reach a fair to good level, which is quite reasonable for this kind of problematic hall. LF80 (lateral energy fraction) shows the spatial configuration which fails in all simulations because it should be > 25 for symphonic music.
By keeping the lower side walls on the tribunes untreated, space was achieved for the portable seating area on the sports field. However, it was not possible to provide this for the tribune seats. In order to provide stage support, removable reflectors made of timber and medium-density fiber boards were proposed for use only during music activities, as seen in Figure 7
. These reflectors will be 2.5 m wide and 4 m high.