Handbook for Sound Engineers

1398 Chapter 36

and particularly in distributed loudspeaker systems, a
considerable number of early reflections and sound
arrivals will occur at a given listening position. These
can provide a useful bridging effect (sequential masking)
which can extend the useful arrival time to perhaps
50 ms. The way in which single or discrete reflections
affect intelligibility has been studied by a number of
researchers—perhaps the best known being Haas.

Haas found that under certain conditions, delayed
sounds (reflections) arriving after an initial direct sound
could in fact be louder than the direct sound without
affecting the apparent localization of the source. This is
often termed the Haas effect. Haas also found that later
arriving sounds may or may not be perceived as echoes
depending on their delay time and relative level. These
findings are of significant importance to sound system
design and enable, for example, delayed infill loud-
speakers to be used to aid intelligibility in many applica-
tions ranging from balcony infills in auditoria and pew
back systems in churches to large venue rear fill loud-
speakers. If the acoustic conditions allow, then
improved intelligibility and sound clarity can be
achieved without loss of localization.

Fig. 36-24 presents a set of echo disturbance curves
produced by Haas and shows the sensitivity to distur-
bance by echoes or secondary sounds at various levels
and delay times.

Fig. 36-25, after Meyer and Shodder, shows a curve
of echo perception for various delay times and levels
(dotted curve) and indicates that delayed sounds become
readily discernible at delays in excess of 35 ms (e.g., at
50 ms delay), a single reflection or secondary signal has
to be more than 10 dB lower before it becomes imper-
ceptible and has to be more than 20 dB lower at 100 ms.
The solid curve in Fig. 36-25 shows when a delayed
sound will be perceived as a separate sound source and
ceases to be integrated with the direct sound.

Although potentially annoying, echoes may not

degrade intelligibility as much as is generally thought.

Fig. 36-26, based on work by Peutz, shows the reduction

in %Alcons caused by discrete sound arrivals or echoes.

The curve starts at just under 2% as this was the residual

loss due to the particular talker and listener group taking

part in the experiment. As the figure shows, the single

reflections typically only caused an additional loss of

around 2–3%.

However, typically more complex systems operating

in reverberant spaces can often give rise to the creation

of groups of late reflections which, anecdotally at least,

Figure 36-23. Intelligibility versus distance for a four meter
line array in a reverberant church.

STI vs Distance for Line Array in 4 s RT Church

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

2 4 6 8 101517202224262830

Distance–m

STI

Figure 36-24. Echo disturbance as a function of delay and

level (after Haas).

Figure 36-25. Echo perception as a function of delay time

and level (after Meyer and Shodder).

Figure 36-26. Effect of echoes on %Alcons (after Peutz).

Time interval (ms) between primary and secondary sounds

% listeners disturbed

Time–ms

dB

0 10 20 30 40 50 60 70 80 90 100

0 1 2 3 4 5 6 7

Delay time–ms

%ALcons for reflections at equal level to direct

sound for various values of delay time