Introduction to Acoustics

the duplex theory of sound localization, proposed by
Lord Rayleigh [13.149]. However, complex sounds, containing
only high frequencies (above 1500 Hz), can be
localized on the basis of interaural time delays, provided
that they have an appropriate temporal structure.
For example, a single click can be localized in this
way no matter what its frequency content. Periodic
sounds containing only high-frequency harmonics can
also be localized on the basis of interaural time differences,
provided that the envelope repetition rate (usually
equal to the fundamental frequency) is below about
600 Hz [13.150, 151]. Since most of the sounds we encounter
in everyday life are complex, and have repetition
rates below 600 Hz, interaural time differences can be
used for localization in most listening situations.
13.7.2 The Role of the Pinna and Torso
Binaural cues are not sufficient to account for all aspects
of sound localization. For example, an interaural
time or intensity difference will not indicate whether
a sound is coming from in front or behind, or above or
below, but such judgments can clearly be made. Also,
under some conditions localization with one ear can
be as accurate as with two. It has been shown that reflections
of sounds from the pinnae and torso play an
important role in sound localization [13.152, 153]. The
spectra of sounds entering the ear are modified by these
reflections in a way which depends upon the direction of
the sound source. This direction-dependent filtering provides
cues for sound source location. The spectral cues
are important not just in providing information about
the direction of sound sources, but also in enabling us
to judge whether a sound comes from within the head
or from the outside world. The pinnae alter the sound
spectrum primarily at high frequencies. Only when the
13.8 Auditory Scene Analysis
It is hardly ever the case that the sound reaching our ears
comes from a single source. Usually the sound arises
from several different sources. However, usually we are
able to decompose the mixture of sounds and to perceive
each source separately. An auditory object can be defined
as the percept of a group of successive and/or simultaneous
sound elements as a coherent whole, appearing to
emanate from a single source.
As discussed earlier, the peripheral auditory system
acts as a frequency analyzer, separating the different fre-
Psychoacoustics 13.8 Auditory Scene Analysis 485
wavelength of the sound is comparable with or smaller
than the dimensions of the pinnae is the spectrum significantly
affected. This occurs mostly above about 6 kHz.
Thus, cues provided by the pinnae are most effective for
broadband high-frequency sounds. However, reflections
from other structures, such as the shoulders, result in
spectral changes at lower frequencies, and these may be
important for front–back discrimination [13.154].
13.7.3 The Precedence Effect
In everyday conditions the sound from a given source
reaches the ears by many different paths. Some of it arrives
via a direct path, but a great deal may only reach
the ears after reflections from one or more surfaces.
However, listeners are not normally aware of these reflections,
and the reflections do not markedly impair the
ability to localize sound sources. The reason for this
seems to lie in a phenomenon known as the precedence
effect [13.155, 156]; for a review, see [13.157]. When
several sounds reach the ears in close succession (i. e.,
the direct sound and its reflections) the sounds are perceptually
fused into a single sound (an effect called echo
suppression), and the location of the total sound is primarily
determined by the location of the first (direct)
sound (the precedence effect). Thus the reflections have
little influence on the perception of direction. Furthermore,
there is little awareness of the reflections, although
they may influence the timbre and loudness of the sound.
The precedence effect only occurs for sounds of
a discontinuous or transient character, such as speech
or music, and it can break down if the reflections have
a level 10 dB or more above that of the direct sound.
However, in normal conditions the precedence effect
plays an important role in the localization and identification
of sounds in reverberant conditions.
quency components in a complex sound. Somewhere in
the brain, the internal representations of these frequency
components have to be assigned to their appropriate
sources. If the input comes from two sources, A and
B, then the frequency components must be split into
two groups; the components emanating from source A
should be assigned to one source and the components
emanating from source B should be assigned to another.
The process of doing this is often called perceptual
grouping. It is also given the name auditory scene anal-
Part D 13.8