Handbook for Sound Engineers

Designing for Speech Intelligibility 1399

would appear to be rather more detrimental. Fig. 36-27
shows the ETC measured on the stage of a 1000 seat
concert hall auditorium. A small group of prominent late
reflections is clearly visible.

The reflections arrive some 120 ms after the direct
sound and are less than 0.5 dB lower and would there-
fore be expected to be a significant problem. The cause
was sound from the center cluster rebounding off the
acoustically untreated rear wall and returning to the
stage. This was not only clearly audible but also
extremely annoying to anyone using the system when
speaking from the stage—the coverage and intelligibil-
ity throughout the audience area, however, were
extremely good. The problem, although clearly caused
by the sound system, was in fact, not the fault of the sys-
tem, but rather the lack of appropriate acoustic treatment
on the rear wall. Sound from the cluster had to strike the
wall in order to cover the rear rows of seating. Although
this was released at the design stage and appropriate
treatment arranged, in the event this was not installed
and an extremely annoying echo resulted. (Later instal-
lation of the specified treatment solved the problem,
which shows how important it is to properly integrate
systems and acoustics.)

Another interesting problem found in the same audi-

torium during initial setting up of the system is shown in

Fig. 36-28. Again a group of late reflections is clearly

visible. A strong reflection occurred 42 ms after the

direct sound just 1.9 dB down and the later group

arrived 191 ms after the direct and 4.5 dB down.

Perhaps surprisingly, the effect of these reflections

was not to create a distinct echo but rather to cause a

general loss of intelligibility and blurring of the sound.

In other nearby seats, the intelligibility was good and

measured 0.70 STI but in the seats where the intelligibil-

ity was poor the STI was 0.53. Although significantly

lower than 0.7, a value of 0.53 would still appear to be

too high in relation to the subjective impression

obtained. However, Houtgast and Steeneken specifically

warn against the use of STI for assessing situations with

obvious echoes or strong reflections. Identifying the

problem however, would not have been possible without

the ability to see the ETC.

36.9 Uniformity of Coverage

It is essential when designing systems to work in noisy

and/or reverberant spaces to insure that the direct sound

level is as uniform as practical. For example, while a

6 dB variation (±3 dB) may be acceptable under good

acoustic conditions, such a variation in a reverberant

space can lead to intelligibility variations of 20–40%. A

40% degradation of clarity under such conditions is

usually unacceptable. For the case of noise alone, the

variation would be at least a 20% reduction in potential

intelligibility—though this will be dependent upon the

spectrum of the noise. The off-axis performance of a

selected loudspeaker is therefore of critical impor-

tance—a smooth and well-controlled response being a

highly desirable feature.

Figure 36-27. Impulse response and ETC of late reflection
from auditorium rear wall to stage.

Direct F = 18 ms

Reflection ¾Ms

Impulse Response

D R

Figure 36-28. ETC showing late reflections in auditorium

causing blurring of sound and loss of intelligibility.

Side balcony 1500 Hz note reflections @70 ms

R (^1) R 2
R 1 = 42 ms 1.7 dB R 2 = 42 ms 4.5 dB
Time–s
D