Handbook for Sound Engineers

1394 Chapter 36

high (i.e., a low Q device is used), then the “ba” sound
will have only decayed by approximately 18 dB and will
completely mask the “ck” sound by 8 dB to 13 dB. It
will therefore not be possible to understand the word
back or distinguish it from similar words such as bat,
bad, bath, or bass since the all important consonant
region will be lost. However, when used in the context
of a sentence or phrase, the word may well be worked
out by the listener from the context. Further increasing
the reverberation time (or reverberant level) will further
increase the degree of masking.

Not all reverberation, however, should necessarily be
considered to be a bad thing, a degree of reverberation is
essential to aid speech transmission and to aid the talker
by returning some of the sound energy back to him or
her. This enables subconscious self-monitoring of their
speech signal to occur and so feed back information
about the room and projected level. The room reverbera-
tion and early reflections will not only increase the per-
ceived loudness of the speech, thereby acting to reduce
the vocal effort and potential fatigue for the talker, but
also provide a more subjectively acceptable atmosphere
for the listeners. (No one would want to live in an
anechoic chamber.) However, as we have seen the bal-
ance between too much or not enough reverberation is a
relatively fine one.

The sound field in a large space can be highly com-
plex. Statistically, it can be divided into two basic com-
ponents, the direct field and the reverberant field.
However, from the point of view of subjective impres-
sion and speech intelligibility the sound field needs to
be further subdivided to produce four distinct compo-
nents. These are:

1. Direct Sound—that directly from source to listener.


2. Early Reflections—arriving at the listener approxi-
   mately 35–50 ms.


3. Late Reflections—arriving at the listener approxi-
   mately 50–100 ms later (though discrete reflections
   can also be later than this).


4. Reverberation—high density of reflections arriving
   after approximately 100 ms.

Fig. 36-18 summarizes the sound field components
discussed above.

To the above list one could also add “Early Early”
reflections—those occurring within 1–5 ms. (If specular
in nature, these generally cause comb filtering and
sound coloration to occur. Reflections of 1–2 ms are
particularly troublesome as they can cause deep notches
in the frequency response to occur around 2 kHz and

thereby reduce intelligibility by attenuating the primary

speech intelligibility frequency region.)

Opinion as to how the direct sound and early reflec-

tions integrate is currently somewhat divided. Many

believe that reflections occurring up to around 35–50 ms

after the direct sound fully integrates with it, provided

that they have a similar spectrum. This causes an

increase in perceived loudness to occur, which under

noisy conditions can increase the effective SNR and

hence intelligibility. Under quieter listening conditions,

however, the case is not quite so clear, with factors

including spectral content and direction of reflection

becoming increasing important. Equally some research

suggests that the integration time may be frequency

dependent but generally around 35 ms for speech sig-

nals. However, there is general agreement that later

arriving reflections (>50 ms) act such as to degrade

intelligibility with increasing effect as the arrival time

delay increases.

Sound arriving after approximately 100 ms generally

signals the start of the reverberant field though strong

discrete reflections arriving after 60 ms or so will be

heard as discrete echoes. It is the ratio of direct + early

reflections to late reflections and reverberation that

determines the potential intelligibility in a reverberant

space (assuming that other effects such as background

noise and frequency response considerations are

neglected). As a rule, positive ratios are desirable but

rarely achieved in reality, though there are exceptions.

This is demonstrated in Figs. 36-19 and 36-20. Fig.

36-19 shows the energy time curve (ETC) sound arrival

analysis for a highly directional (high Q) loudspeaker in

a large reverberant church (RT 60 =2.7s at 2kHz). The

D/R ratio at the measuring position (approximately^2 / 3

way back) is 8.7 dB resulting in a high degree of intelli-

gibility. Other intelligibility measures taken from the

Figure 36-18. Sound field components.

DIRECT

SOUND

EARLY

REFLECTIONS

LATE

REFLECTIONS

REVERBERATION

Time Seconds

LEVEL

dB