Page 234 - Master Handbook of Acoustics

The 50 rooms of the Jackson-Leventhall study were of varying sizes, shapes, and degree of
3
3
furnishing. The sizes varied from 880 to 2,680 ft , averaging 1,550 ft . An optimum reverberation time
for speech for rooms of this size is about 0.3 second (see Fig. 11-15). Only those living rooms near
the lower limit approach this, and in them we would expect to find much heavy carpet and overstuffed
furniture. These reverberation measurements tell us little or nothing about the possible presence of
timbral defects. The BBC engineers checked for timbral changes and reported problems in a number
of the living rooms studied.

Artificial Reverberation

Artificial reverberation is considered a necessity for many kinds of music recording projects. Music
recorded in dry (nonreverberant) studios lacks the richness of the room effect contributed by most
acoustical spaces. The addition of artificial reverberation to such recordings is standard practice, and
there is great demand for processors that will provide natural-sounding artificial reverberation at a

reasonable cost.
There are many ways to generate artificial reverberation, but the challenge is finding that method
that mimics actual acoustical spaces and does not introduce frequency-response aberrations into the
signal. Historically, a dedicated reverberation chamber was employed. The program is played over a
loudspeaker in this room, picked up by a microphone, and the reverberated signal mixed back into the
original in the amount to achieve the desired effect. Small reverberation chambers are afflicted with

serious timbral problems because of widely spaced modes. Large chambers are expensive. Even
though reverberation chambers have certain desirable qualities, the problems outweigh the
advantages and they are now a thing of the past.
Most artificial reverberation is created digitally using digital hardware or software programs. The
principle of digital reverberation is illustrated in the signal flow schematic of Fig. 11-19. The input

signal is delayed, and a portion of the delayed signal is fed back and mixed with the incoming signal,
the mixture being delayed again, and so on. This replicates the action of sound rays reflecting from
many room surfaces with an attenuation at each reflection.

FIGURE 11-19 A simple digital reverberation algorithm using a delay line and feedback.

Schroeder found that approximately 1,000 echoes per second are required to avoid a flutter effect
and to sound natural to the ear. With a 40-msec delay, only 1/0.04 = 25 echoes are produced each