Assignment Two – Audio Systems Design and Evaluation

Audio Systems and Measurement – DESC9090, Architecture Design and Planning,

University of Sydney, Semester Two 2018

Introduction
Emerging audio systems in stadiums and auditoriums in the last century have been met with
several technological and acoustical limitations. In the past decades, high quality digital audio
systems, array speakers and computer aid simulations created greater flexibility where high-
quality audio systems were in demand. Nowadays, array speakers with their cylindrical sound
propagation are more popular in large halls due to lower sound attenuation by distance (3dB
attenuation with doubling the distance) and their ability to produce a wide range of frequency
bands and physical coverages.
This assignment is allocated to audio system design, simulation and evaluation of the provided
auditorium cad (picture 1) through the use of EASE acoustic (software). The design target has
been based on mapping direct sound for obtaining a coverage uniformity across seating areas
from 125Hz to 8kHz according to ANSI/InfoComm Standard A102.01-2017 (6dB range
coverage) [1] and using the Aura calculation module for evaluating the calculated spectral
balance based on the AVIXA standard (A103.01:201X) over the seats [2]. In addition, the
minimum amount of the Total SPL has been requested to be 110 dBA (lower than FIFA
recommendation) and STI of 0.75 for males [5].
A sound system has been designed by the author for providing the results as close as possible
to the target demands. Eight QSC array speakers (appendix 1) have been used for coverage of
the entire auditorium and the low frequency direct levels for octave band frequencies (31.5Hz
& 63Hz) have been calculated by EASE Focus 3 (version 3.1.5) [4].

Figure 1. This picture shows the auditorium design with seats and speaker array locations.
System design
Using all speakers over the front stage can create the best sense of localization but has some
limitations regarding the coverage of all audience areas in different distances and elevations
[7,8]. By considering the dimension of auditorium (~58mX80m), a powerful speaker system
will be needed to cover the desired angles and distances. Overlapping of acoustical pressure of
multiple array speakers can create interference patterns that can affect the level of sound by
constructive and destructive summations due to phase differences between similar signals. The
patterns of level differences can cover up larger areas for lower frequencies on overlapping
regions due to longer wavelengths while the coverage of the mid frequencies can be more
homogenized [8]. The high frequencies (>4K) are under greater influence of dissipation by
distance (atmospheric absorption). As there is no balcony in the auditorium construction, filling
all audience areas from the overhead of the stage can be possible with array speakers. Seven
full range QSC array speakers with horizontal angle of 140 degrees have been used in a
symmetrical arrangement (3x left, center, 3x right) in the correspondent angles and distances
for the optimum coverage of direct sound levels as close as possible to the target (6dB
deviation). Such system resembles the multichannel system with three apparent origins as
suggested by ANSI/INFOCOMM A102.01:2017 for a uniform coverage [1]. The center array
has been adjusted in the number of elements and angles (table one) so that it could cover up
the floor and central upper tiers. The left and right arrays with 7ms calculated delay in respect
to the center speaker are adjusted in the number of elements and angles (vertical and horizontal)
for creating an optimum coverage over the desired areas. These speakers are installed as close
as possible in respect to each other to reduce the uneven patterns caused by delays and
interferences. As high frequency dissipation is higher for longer distances (upper tiers), the side
arrays (L2,3 & R2,3) are adjusted and aimed to compensate for the sound power dissipation in
these areas. For reducing the low frequency interference between the arrays and enhancing the
sound quality (decreasing the THD in high level reproduction), a bass array (table 1) has been
installed behind the main full range central array and low frequencies under 250Hz are
attenuated for speaker arrays (L2, L, C, R, R2) by a 3rd order roll off filter. The central bass
system (KLA181) can reproduce frequencies from as low as 33Hz (with higher purity and
lower THD) to up to 250Hz. The loudspeaker arrangement (L3, Bass, R3) has been designed
to cover up the lower frequencies (<125Hz) for all areas. The details of the typical electro
acoustical specifications of the speakers have been shown in appendix 1 and the arrangement
of arrays has been tabulated in table 1.
Array Model X(m) Y(m) Z(m) Angle Delay Number EQ Configuration
Name (QSC) Hor Time(ms) of Filter (Degree)
Elements
Left3 WL2102 11.5 0.9 18.62 -220 7 4 None Large Frame (WL2102)
(4-0-2-4)
Left2 WL2102 9.5 0.9 18.62 -199 7 6 HP Large Frame (WL2102)
(250Hz) (0-0-0-4-4-4)
3rd Order
Left WL2102 7.76 0.9 18.62 180 7 16 HP Large Frame (WL2102)
(250Hz) (0-0-0-4-0-0-6-6-6-10-10-6-6-6-6-6)
3rd Order
Centre WL2102 0 0.9 18.62 180 0 14 HP Large Frame (WL2102)
(250Hz) (4-0-2-4-6-8-10-10-10-10-10-10-10-10)
3rd Order
Bass KLA181 -2 18.62 180 0 5 None KLA181 (0-0-0-0-0)
Table 1. Loudspeaker arrays configuration details. The details of right channels are as same
the left channels.
Calculations
The calculations have been done in four different steps.
1. Direct sound level (dB SPL Z weighted)
2. Aura (Particles-Intermediate Resolution and Length-Standard, Fast), not considering
background noise and signal masking.
3. Aura (Particles-Intermediate Resolution and Length-Standard, Fast) considering
background noise and signal masking.
4. Low frequency direct sound calculation (EASE Focus 3).

Coverage Uniformity
As depicted in figure 2, the standard deviation of the octave bands direct levels has shown a
good compatibility with the 6dB coverage uniformity standard according to
ANSI/INFOCOMM A102.01:2017 [1, 9]. Table 2 has shown the percentage of coverage in
and out of the 6dB coverage target while figure 3 has depicted the coverage maps for all
frequency bands. It has been observed that the areas are not evenly covered due to the
interference effect between the array elements and overlapping regions between the separated
speakers. The data has also shown that the coverage uniformity is frequency dependent and
achieving an even coverage for low frequencies (125Hz - 250Hz) is much harder than high
frequencies due to the interaction of their longer wavelengths between the multiple loud
speaker arrays. In this design the issue of low frequency interference has been overcome by
filtering the low frequencies for main arrays and reproducing the low-end by central sub arrays
and full range side arrays. Line array loudspeakers can attenuate high frequencies due to the
signal summation between the array elements. In this case, the higher number of elements in
the L-C-R array speakers has created more attenuation on high frequencies (>6kHz) for closer
seats but farther seats have received more HF attenuation due to the angles between the array
elements and long-distance atmospheric dissipation [7].
f (Hz) 125 250 500 1000 2000 4000 8000
Average (dB) 107.51 109.26 106.97 106.2 106.72 106.34 101.29
6dB coverage envelop % 85 88 94.22 95.48 93.66 95.41 95.93
8dB coverage envelop % 91.95 92.96 98.18 98.73 97.9 98.93 98.57
12dB coverage envelop % 99.81 99.76 99.99 100 100 100 100

Table 2. The percentage of coverage uniformity for octave band frequencies according to
standard ANSI/INFOCOMM A102.01:2017 [1].

115
110
Level (dB)

105 Figure 2. The octave band average

100 direct response and the statistical
95
90 curves across the auditorium areas.
85
The standard deviation has shown a
max of ± 3dB deviation across the
frequency band compatible to standard
Frequency (Hz)
ANSI/INFOCOMM A102.01:2017 for
Average Avg + StdDev Avg - StdDev
6dB direct level deviation over the
areas [1]
Maximum Minimum
Figure 3. Graphical and statistical demonstration of coverage uniformity across the
auditorium calculated for octave frequency bands between 125Hz to 8kHz.
Spectral Balance
According to the standard AVIXA A103.01:201X, DS1 2018 [2], the spectral balance
frequency response has been calculated by the power averaging of the direct octave frequency
response amplitudes (calculated by Aura) over the 12 seats. The resulting normalized frequency
response (@1KHz) has been drawn on the standard curves (SB-A) [2] and has shown a good
compatibility with the target except at 200Hz which has been overestimated by 0.8 dB (figure
4). By considering the low-frequency response (35Hz lower limit) and the spectral balance
frequency response, the designed sound system shall be classified as a full bandwidth audio
system [2].

8
6
4
Level (dB)

2
0
-2
-4
-6
-8
Frequency (Hz)

Power Average Response Upper Limit (SB-A) Lower Limit(SB-A)

Figure 4: The power averaged response of the 1 octave band response for the 12 seat sound
levels has shown a good compatibility to the standard criterion SB-A [2].

Frequency Response
According to figure 5 (left), the average frequency response has shown a good linearity (±1dB
deviation) from 300Hz to 7KHz. The drop of high frequency can be in relation to the
interference between array units and the overall dissipation of high frequency by air. The
accumulation of low frequencies due to acoustical characteristics such as reverberation time
can result in higher levels on low frequencies. The low-end response declination is due to
electromechanical characteristics of the speaker drivers (sub woofers). The individual response
of all seats (figure 5. right) have shown ±3B deviation from the average which is compatible
with the standard ANSI/INFOCOMM A102.01:2017 [1]. According to AVIXA
A103.01:201X, DS1 2018 [2], this system can be classified as full bandwidth as its response
with a 1 kHz reference is greater than or equal to -5 dB at 40 Hz and is greater than -10 dB at
12.5 kHz that makes it appropriate for music and speech. A flat frequency response can enhance
the naturality and definition of sound, but it is better to provide a flat frequency response by a
high quality and full range speaker system with minor EQ in levels. Using EQ can be suitable
for adjusting the small deviations but high-level compensations, especially amplifying low
frequencies in some speaker drivers, can result in more distortion and coloration. Emphasizing
some frequencies such as 500 Hz to 4000Hz can improve speech intelligibility for moderate
levels due to higher weighting factors for these frequencies [3] but emphasizing low
frequencies may result in higher auditory masking for important mid frequencies which reduces
the intelligibility.
110

6 105

Level (dB)
4 100
2
95
Level (dB)

0
90
-2

-4
Freqency (Hz)
-6
Seat1 Seat2 Seat3 Seat4
-8 Seat5 Seat6 Seat7 Seat8
Frequency (Hz) Seat9 Seat10 Seat11 Seat12

Figure 5. 1/3 octave band average frequency response (left) and the overlaid individual
frequency responses of the 12 seats (right) calculated by Aura.

Low frequency augmentation

As shown in figure 6, low frequency calculation has been done for 3 arrays (L3-Bass-R3) by
EASE Focus V3 [3]. The low frequencies for other speaker arrays have been cut (HP-3rd order
filter) for minimizing the low frequency interaction between the arrays and enhancing the
quality and definition of low-end coverage reproduced by central sub woofers. In addition, the
horizontal angle of array speakers and the distance between them has been adjusted for
minimum overlapping and interaction for creating the optimum low frequency coverage. The
statistical data has shown that the coverage uniformity (6dB envelop) has been achieved for
more than 90% of areas in both low octave band frequencies (31.5 Hz & 63Hz). Figure 6 (right)
has shown more coverage anomality due to the interaction between the arrays and summation
on overlapping areas but for frequencies lower than 45Hz (left picture), the response has solely
been created by the central Bass array speaker (KLA181) and shows the regular attenuation by
distance.
According to EASE Focus 3 [4], using 5 elements of KLA181 in one array can cause
overweighting issue. Although author has used them in an integrated array (compatible to
EASE 4.4), in a real-world design the elements can be hung by two separated rigging systems
for avoiding the overweighting issue.
 Figure 6. low frequency direct level (Z-
weighted) calculation for 31.5Hz (upper left),
63Hz (upper right) and the calculated low
frequency response (<125Hz) for five
locations has been shown in the left picture.
The frequency response curves have shown ±
3dB tolerance for coverage uniformity in the
measured areas that is compatible to the FIFA
standard [5].

SPL (Total)
The SPL (total) has been calculated through the summation of the levels of all A-weighted
octave frequency bands (125Hz – 8kHz) according to the FIFA suggestions [5]. The results
(table 3) have shown a good compliance by the FIFA target (110 + 6dB) with about 2 dB (<
±3dB) deviation in SPL level over the seating areas (12 seats).

Seats 1 2 3 4 5 6 7 8 9 10 11 12
SPL (Total) dBA 116.6 118.1 114.6 118.1 118 115.2 116.8 116.3 116.8 115.9 114.4 115

Table 3. A-weighted SPL (total) of direct sound (Aura calculation) over the 12 seats.
 Acoustical Characteristics

• Reverberation Time (RT)

Figure 7 has shown the calculated average reverberation time for single (center speaker) and
multiple (all speakers) source by Aura module. The results have shown that using multiple
speakers has overestimated the RT for mid and high frequencies and underestimated the RT
for low frequencies. As the RT can be changed slightly by the number and position of the
loudspeakers, using all speakers for measuring acoustical parameters such as RT cannot be
suitable for reporting the metrics. In addition, using simultaneous multiple speakers for RT
measurements has not been standardized [6].
The T30 curves also have shown longer reverberation time (up to 2.8s) for the lower
frequencies due to lower absorption coefficient in these bands that can accumulate more sound
energy in these bands and reduce the STI by affecting the MTIs. Furthermore, declining the
RT for high frequencies is due to higher absorption coefficient for these frequencies in the
auditorium.

2.8

2.3
RT (s)

1.8

1.3

0.8

Frequency (Hz)

T20 All Sp T30 All Sp T20 C Sp T30 C Sp

Figure 7. The averaged auditorium reverberation time (T20, T30) calculated by ray tracing
module (Aura) for all speakers (All Sp) and center speaker (C Sp).

• Speech Intelligibility (STI) and Clarity Index (C50 and C80)

The calculated STI with and without signal masking (table 4) has shown an overall lower
amount of STI with respect to the target (0.75) for most of the seats but the sound system has
provided sufficient speech intelligibility (>0.55) in all areas [1]. The clarity index curves (figure
9) have shown a good music clarity C80 for frequencies higher than 315Hz and speech clarity
(C50) for frequencies higher than 400Hz that are compatible with the calculated STI amounts.
The results also have shown that the STI amounts were not changed by the background noise
as the signal to noise has been higher than 15dB [3] but they have been reduced by accounting
the signal masking effect [3]. The STI distribution is not the same for all areas and has shown
slight variation due to the level of signal, reverberation time and auditory masking [3]. The
highest amounts have been observed for seats #1 and #2 (floor) and the lowest amount has been
observed for seat #4. The lower STI for seat #4 can be predicted from echo speech graphs
 (figure 8) as the level of delayed spikes between 750 ms to 1000 ms are higher for seat #4.
Such late echoes can negatively impact intelligibility.

Figure 8. Echo Speech comparison between seats #1 and #4 at 1kHz.

16
14
12 Figure 9. The averaged auditorium
10 clarity index (C50 and C80) calculated
dB

8
6 for 12 seats for one speaker (center).
4 Auditorium has provided high clarity
2
0 index for high frequencies that is
satisfy with the lower RT for those
frequencies and low clarity for low
frequencies due to high RT at low
Frequency (Hz)
frequencies.
C50 C80

Seats 1 2 3 4 5 6 7 8 9 10 11 12
STI (Male) plus
Noise and Masking 0.645 0.657 0.608 0.585 0.626 0.602 0.606 0.591 0.572 0.586 0.569 0.589
Verification Good Good Good Fair Good Good Good Fair Fair Fair Fair Fair
STI (Male) 0.748 0.774 0.702 0.66 0.725 0.694 0.71 0.682 0.651 0.675 0.657 0.679
Verification Good Excellent Good Good Good Good Good Good Good Good Good Good

Table 4. The auditorium speech intelligibility index (STI) for individual seats. The results are
calculated by considering the background noise and signal masking in comparison to none of
them.

Conclusion
A symmetrical audio system consisting of 8 array speakers (QSC- types WL2102 and
KLA181) has been designed for the nominated auditorium in the EASE 4.4 platform. The
resulting data derived from direct sound mapping have shown a coverage uniformity from 85%
(125Hz) to 95.93% of areas of interest comply to the FIFA [5] standard for 6 dB coverage
uniformity. It has been observed that covering all areas in this range cannot be achieved due to
the acoustical interactions between the array elements and adjacent array speakers.
Furthermore, the calculation results from Aura mapping module have shown a good
compatibility of spectral balance with the SB-A [2] curves with a small deviation in 200Hz.
The data derived from low frequency augmentation and the calculated frequency responses
have shown that the system can reproduce a full range frequency band suitable for speech and
music. The calculated A-weighted SPL (total) [5] has shown 5 dB to 8dB higher SPL in
comparison to the assignment’s target (110dBA) for all seats. Finally, the designed audio
system has shown a Fair to Good (>0.55) amount of STIs in presence of auditory masking for
all seats [5]. The STI has shown a good uniformity over the audience areas and the small
deviations are due to a combination of reverberation time, audio masking and levels.

References

[1] American National Standards Institute. (2017). Audio Coverage Uniformity in Listener
Areas (A 102.01). Fairfax, VA: InfoComm International.

[2] Audiovisual and Integrated Experience Association. (2018). Spectral Balance of Sound
Systems in Listener Areas (A103.01:201X). Fairfax, VA: AVIXA.

[3] British Standard. (1998). Objective rating of speech intelligibility by speech transmission
index: part 16: (IEC 60268-16:1998). London, 389 Chiswick High Road: BSI.
[4] Ease Focus 3 [Computer software]. (2018). Retrieved from http://focus.afmg.eu/
[5] Fédération Internationale de Football Association. (2014). Technical Specification for PA
System (50.20.20.50). Switzerland: FIFA.

[6] ISO, E. (2009). 3382-1, 2009, “Acoustics Measurement of Room Acoustic Parameters
Part 1: Performance Spaces,”. International Organization for Standardization, Brussels,
Belgium.
[7] Miranda,L (2018). Audio System Design Principles - DESC9090. Audio Systems &
Measurement -week 8.
[8] Miranda,L (2018). Principles of Acoustic Modelling - DESC9090. Audio Systems &
Measurement -week 10.
[9] Swanson, P. The Audio Coverage Uniformity Standard [Ebook] (pp. 62-63). Retrieved
from http://av.net.au/contents/issue_11/audio_coverage_uniformity.pdf
Appendix 1.

QSC Speakers’ specifications:

Retrived from https://www.qsc.com/