Recording - February 2016 - 68

By Paul J. Stamler

Part 14: Digital Recording 1
We think of digital audio primarily as a way to manipulate sounds, but
that's not how it started out. Originally, the idea was to make audio signals
less vulnerable to corruption.
Like many things in audio, digital recording came out of research done by
The Phone Company. I capitalized those words because, way back when, in
most of the USA there was only one phone company: AT&T/Bell Telephone.
In fact, the first practical application of the technology that became digital
audio recording was a system called SIGSALY, a private (and very secure)
communication system which connected Winston Churchill and Franklin D.
Roosevelt during World War II. Back then, digital encoding of audio signals
was pretty exotic stuff; now it's become universal.
Why did that happen, and why is digital recording useful, anyway? To
know that, we need to take a (massively simplified) look at how it works.

VOLTAGEg

TIMEg
Sampling and numbers
Take a look at Figure 1, a very short snippet of an audio waveform. It's
actually a plot of voltage (the Y-axis) versus time (the X-axis). This is the sort of
waveform you might find at the output of a mic preamp, and you can make it
into a useful recording in several ways.
You could cut a groove shaped like the waveform in a wax master disc
and use that (via intermediate steps) to press records, as was done during the
78 rpm era. Or you could use the waveform to vary the amount of magnetism in the coil of a tape recorder's record head, causing the orientation of
tiny particles of iron oxide on a strip of plastic to vary in a way that mimics
the ups and downs of the waveform. The groove cut in the master disc or the
magnetic pattern in the oxide layer will be more or less analogous to the ups
and down of the waveform, so we call this analog recording, and for the first
century or so of the recording industry that was pretty much all there was.
The shape of the groove would be traced by a needle, or the orientation of
the tiny oxide particles would be sensed by a tape recorder's playback head,
and a voltage would be produced that followed the voltage of the original
waveform-more or less.
68

RECORDING February 2016

There were several reasons for the "more or
less" qualification. The big one was noise;
when the needle played back the replicated
groove on a phonograph record, it picked up
random fluctuations in the signal, caused by
the material the record was pressed from; likewise, the tape recorder's playback head
picked up random fluctuations in the tiny particles of oxide. These random fluctuations
weren't in the original signal, but were added
by the storage medium (shellac or vinyl discs,
magnetic tape). They were, in fact, noise.
These recording processes were also nonlinear,
which I've mentioned is an engineer's way of saying that they delivered a distorted version of the
original signal.
The same problems (noise and distortion) would
affect a signal as it passed down a telephone line
or long-distance microwave link, so it was natural
for the phone company to seek a remedy.
Another way
The engineers at Bell figured out that if,
instead of transmitting a signal, you could
describe it, and transmit the description, you'd
do an end run around the imperfections of
transmission-or, if you recorded and played
back the description, the imperfections of
recording media. The way they chose to
describe the signal was with numbers, and to
do that they had to sample the signal, which
meant measuring it at regular intervals and
deriving a number which represented the voltage at the moment the signal was sampled.
The device that does this is an analog-to-digital
converter, or A/D converter.
Looking at the snippet in Figure 1, the first
sample might be low, corresponding to the
number 269. (269 of what? In this case, 269
of the tiny slices that make up the full signal.
We'll get back to that.)
At the next sample, the signal might be
1,209 slices high; a moment later, it might be
5,414, then it might go down to 1,332. If you
could transmit those numbers (and a bunch
more like them) down a wire to a receiver
which would register the numbers and produce
a voltage of corresponding level (a digital-toanalog converter or D/A converter), then you

Table of Contents for the Digital Edition of Recording - February 2016