Handbook for Sound Engineers

Delay 809

A tapped delay a shown in Fig. 24-9 is suitable for
creating early reflections. Delays T 1 ... TN are unequal
in length and in the range of 10 ms to 30 ms with ampli-
tudes set by g 1 ... gN as appropriate for the character of
the room. In Fig. 24-10, a reticulating path is provided
via gf that produces the exponentially decaying portion.
This could simply be fed into the start of the reflection
generator, but a more satisfactory result is obtained by
using a separate decay section, where the delay taps are
set more densely and may be over a wider range, typi-
cally between 5 ms and 100 ms. The delay tap times
should be chosen to not be harmonic products of each
other to minimize the buildup of standing waves and
comb filters. Any gain product in the reticulating path
must be less than one, otherwise, the sound will expo-
nentially increase until distortion occurs.

24.2.4 Delay-Based Effects

Flange is an audio effect caused by mixing an original
(dry) copy of a sound with a delayed (wet) copy. The
amount of delay is varied over time, creating a varying
pattern of comb filters that sweep up and down through
the audio spectrum. Chorus is used to make one voice or
instrument sound like many and has the same topogra-
phy as a flanger, but with longer delays.

24.3 Implementations

The implementation of a delay requires some means of

storing the signal and then releasing it after a controlled

period of time. This can be done either by storing a con-

tinuous record of the sound or by breaking it up into

samples that are stored separately. Some preparation of

the signal is usually required to make it compatible with

the chosen storage medium and method and may

involve some postprocessing to restore the stored signal

to a usable form.

24.3.1 Small Delays

Small delays may be realized using the phase shift char-

acteristics of an all-pass filter. Such a circuit, illustrated

in Fig. 24-11, is limited to delays of the order of less

than a wavelength of the highest frequency. These sec-

tions may be chained together to produce longer delays

but become impractical for delays longer than a few

milliseconds. This method is sometimes used in active

crossover systems for a loudspeaker. The delays needed

to time-align drivers within a cabinet are small and

fixed, and the circuit may be easily combined with the

frequency filtering requirements.

24.3.2 Acoustic Delay Methods

One way to implement a long delay is to use the speed

of sound and send the signal to be delayed through a

fixed air path, such as a tube with a loudspeaker at one

end and a microphone at the other. For such a device to

work effectively, the tube must be damped to prevent

internal reflections and have an absorber at one end to

prevent the establishment of standing waves. This type

of system has many disadvantages as the system

becomes very large for any useful delay time. The fre-

quency response changes with tube length due to the

damping material and the signal attenuates as it travels,

meaning that large amounts of gain are required. The

large gain in turn leads to the requirement that the tube

must be mechanically isolated from vibration and out-

Figure 24-9. A delay with multiple taps for creating early

reflections.

Figure 24-10. Delay section with feedback for producing a

decay tail

T 1 ...TN

3

g 1 g 2 gi gN

Delay tap

section

Delay tap

section

gd gf gr

Figure 24-11. An all-pass amplifier having phase shift

proportional to frequency and exhibiting a small amount of

delay.

R

+

C

R

R