Page 96 - Master Handbook of Acoustics

In an auditorium, for example, the room geometry can be designed so that time delays are less than
50 msec and thus fall within the fusion zone. Consider a direct sound with a path length of 50 ft, and
an early reflection with a path length of 75 ft, both reaching a listener. The resulting time delay of 22
msec is well within the fusion zone. Similarly, by limiting the time delay between direct sound and

early reflections to less than 50 msec (a path length difference of about 55 ft), the listener will not
hear the reflections as discrete echoes. If the typical attenuation of reflected sound is accounted for,
differences slightly greater than 50 msec are permissible. Very generally, maximum differences of 50
msec are allowable for speech, while a maximum of 80 msec is allowable for music. Shorter
differences are preferred. As we will see later, the precedence effect can also be used in the design
of live end-dead end control rooms.

The Cocktail-Party Effect

The human auditory system possesses a powerful ability to distinguish between different sounds and
to direct our attention to one sound amid many. This is sometimes called “the cocktail-party effect” or

“auditory scene analysis.” Imagine yourself at a busy party with many talkers and music playing. You
are able to listen to one talker while excluding many other conversations and sounds. If someone
across the room speaks your name, you will be alert to that. There is evidence that musicians and
conductors are highly skilled at this auditory segregation; they can independently follow the sounds of
multiple musical instruments simultaneously.
This ability to distinguish between particular sounds is greatly assisted by our localization

abilities. If the voices of two talkers are played over one loudspeaker, it can be difficult to
differentiate them. However, if two physically separated loudspeakers are set up, and one voice is
played over one speaker, and the other voice is played over the other speaker, it is quite easy to
follow both voices (factors such as relative language, gender, and pitch of the talkers also play a
role). While humans function well at differentiating sources at cocktail parties, electronic signal
processing systems have a more difficult time. This field of signal processing is referred to as source

separation, or blind source separation.

Aural Nonlinearity

When multiple frequencies are input to a linear system, the same frequencies are output. However, the
ear is a nonlinear system. When multiple frequencies are input, the output can contain additional

frequencies. This is a form of distortion that is introduced by the auditory system and it cannot be
measured by ordinary instruments. It is a subjective effect requiring a different approach. This
experiment demonstrates the nonlinearity of the ear and the output of aural harmonics. It can be
performed with a stereo playback system and two audio oscillators. Plug one oscillator into the left
channel and the other into the right channel, and adjust both channels for an equal and comfortable

volume level at some midband frequency. Then tune one oscillator to 23 kHz and the other to 24 kHz
without changing the level settings. With either oscillator alone, nothing is heard because the signal is
outside the range of the ear. However, if the tweeters are good enough, you might hear a distinct 1-
kHz tone.
The 1-kHz tone is the difference between 23 and 24 kHz. The sum, 47 kHz, is another sideband.