1
Computers produce near-perfect sound
QUALITY SPECIFICATIONS
Analog
Digital
Sound/Noise:
65 to 70 dB
85 to 95 dB
Harmonic distortion:
0.05 percent
0.03 percent
Frequency response:
20-25,000 Hz
% or - .5 dB
0-20,000 Hz
% or - .5 dB
Wow & Flutter:
0.05 percent
unmeasurable
By Gary Welch
and Tom Robertson
About once every 25 .years the
public awakens to the fact that an
important step has been made to
improve the recording of music.
In 1877 it was the tinfoil cylin
der that began recording as we
know it today. In 1902, the wax-
process disc was introduced as
the best way to mass produce
recordings, and thus paved the
way for many people to enjoy
recorded music in their homes for
the first time.
Twenty-five years later, people
discovered the tremendous
technological advances that had
taken place in electrical record
ing, and in 1952 the long playing
record (LP) came into its own.
So, in 1977, one would have
expected an exciting innovation
that would turn the music busi
ness upside down again. Digital
recording has done just that.
Early development has shown a
degree of accuracy a leap above
that of any regular recording
equipment.
Since the sound we hear is ac
tually the changing amount of air
pressure on our eardrums
caused by sound waves, conven
tional recording systems adapted
this principal to reproducing the
sound. The height, or amplitude,
of the waves is plotted against a
length of time, and is recorded on
tape, for example, as varying de
nsities of magnetic flux along the
tape. When the tape is played
back, the magnetic tape is read
and converted back into sound.
Similarly, a plastic disc records
sound waves as physical peaks
and valleys cut into a groove of
plastic, and when played the
needle reads the grooves and re
produces the sound.
Digital recording, on the other
hand, takes a different approach.
A computer samples the incom
ing wavelengths of sound very
rapidly and converts the
amplitude of particular points on
the wave into binary numbers.
Binary notation is used in all
computers because the numbers,
being all 0’s and Ts, are easy to
represent electronically as “off”
and “on,” respectively.
The result is a plotted curve
almost identical to the sound
wave. When the tape is played
back, the numbers are converted
into voltage and come out as the
original sound.
The computer is also used to
“average” the curve so that it is
perfectly smooth instead of
slightly jagged. However, the
sampling speed is so high that
any roughness in the curve prob
ably could not be detected by
human ears.
Since the highest pitch a per
son can hear is about 20,000 cy
cles per second (Hertz), the com
puter must sample the sound
waves as least twice as fast to
avoid detectable distortion.
Probably the greatest advan
tage of this new method is that it
reduces noise level and distortion
to virtually insignificant levels.
Noise level is the tape hiss and
other electrical noise that can be
heard when someone turns the
volume up fairly high when listen
ing to a recording. The electrical
noise is caused by the tape pass
ing rapidly over the recording
head. With digital systems, there
is no magnetic tape to cause the
noise, and the result is a very
clean, pure sound.
Distortion in a recording is
caused by innaccuracies in the
waveform, so that when the
music is played it is altered and
irregular. But through the precise
nature of a digital system, ir
regularities are reduced to a very
small amount, and even those
are smoothed out by the “averag
ing” computer.
There are, of course, disadvan
tages and problems in the new
system. The main one is high
cost.
Since the concept, and there
fore the systems, are relatively
new, the software for the com
puters is not fully mass-produced,
therefore making it expensive.
Also, digital editing equipment is
very expensive because it must
be able to track down exact num
bers on the tape, and with 40,000
numbers per .second being rec
orded on tape, the equipment
must be extremely precise.
The best recording systems
now sell for about $150,000, mak
ing digital recording impossible
for some studios. Because of the
cost, digital equipment is now
used in studios only, and is not
commercially available. There
are other less complex systems
that sell for anywhere from
$40,000 to $80,000, which is still
quite expensive.
Another problem lies in the
noncompatibility of the new sys
tems being produced. Differ
ences from system to system
range from sampling rate to
number of binary digits (bits)
used to record the sound, and
they make interchangeability im
possible between many systems.
Most studio digital recorders
use a 16-bit system that samples
at the rate of 40,000 times per
second. This means there is a
need to process 640,000 pieces
of information per second, and to
store one binary number every
1/40,000th of a second, which
leads to a problem of tape satura
tion. The equipment must be very
complex in order to distinguish
between two numbers on the
tape, even if the tape speed is as
high as 30 feet per second.
All significant problems with
digital recording systems, how
ever, can be solved through
greater development and coordi
nation of system components. A
group called the Audio Engineer
ing Society is presently working
to establish standards for profes
sional applications, and better
computer technology and im
proved software is enabling tech
nicians to increase efficiency and
reduce cost at the same time.
Naturally, any system that at
tempts to be as accurate as digi
tal recorders are must have a sys
tem for correcting any possible
errors in transcribing the sound to
the tape. Coding systems have
been set up that arrange bits in
blocks, so that if one bit is acci
dentally dropped the computer
will detect the error and correct it
simply by changing the “1” to a
“0,” or vice versa. In this way the
simplicity of the binary system
works to counteract some of the
complexity of the system as a
whole.
Some commercial recordings
have already been made, and the
system produces such a clean
sound that noise level and distor
tion are virtually nonexistent. Digi
tal recording is indeed the
method of the future; that is, of
course, until someone comes up
with something better.
1111
VOLTAGE
(amplitude)
oooo
l i l -f-H I 1 I I I i I I I I I—t
DIGITAL RECORDING is
based on a system which
samples a sound wave,
left, at a very rapid pace,
and stores on tape a series
of binary numbers that very
closely approximate the
soundwave curve. Keep in
mind that the sampling rate
is so high (about 40,000
per second) that the ir
regularities that exist in this
simplified drawing become
so insignificant as to be in
audible to the human ear.
TIME