Handbook for Sound Engineers

954 Chapter 25

impossible for analog—time-related machinations, for
example, which are typically gruesome in analog.
Tritely, it used to be said that real-deal EQ and
dynamics were the province of analog, being that it has
hitherto been easier and cheaper to achieve
nice-sounding, complex, and flexible phase and
frequency response shaping with a handful of analog
components; this has become less of a black-and-white
proposition though as the size, speed, and power of
digital signal processors have increased and relevant
expertise and ears were applied. Very fine digital EQ,
dynamics, and effects are indeed possible. Suggestion
otherwise “is fightin’ words” and many would suggest
that digital audio processing has now surpassed analog
in all important respects.

Particularly in mixing, switching, and routing there
has been a dramatic bipolar switch over to digital purely
on ease and inexpensiveness of implementation as
appropriate parts became readily available; Fig. 25-124,
a photograph of a couple of LSI ICs and a handful of
support parts, illuminates this blindingly; of course it
could be argued that the same could simply be achieved
in analog with a mere 144 op-amps and 2304 VCAs, but
by whom or why is uncertain.

Digital recording and transmission—including the
most far-reaching domestic example, CD’s are covered
extensively in other sections of this book. That these are
where digits first made their mark on pro audio is hardly
surprising; once the speed of the associated processing
speed and bandwidths were high enough, well-proven
techniques from the communications and computer
world were applied to the problems of storing and
moving the fairly prodigious number of bits digital
audio demands. After all, the major telephone compa-
nies worldwide have been using high-speed digital
streams for decades. Early successes for audio include

the conversion of the BBC’s nationwide radio network

program distribution system to digital in 1971. The turn

of the 80s saw the first few serious digital tape

machines, heralded by the 3M/BBC design; then a little

thing called the PC happened. Hard disk recording

moved from the high-end esoteric to the bedroom studio

and is now both ubiquitous and universal. And very

good. The pro-audio digital revolution is almost

complete. Resistance is futile.

25.17.1 Digital Audio Systems

Fig. 25-126 shows about the simplest example of a DSP

(digital signal processing) system possible. The

processor itself, in old days racks of discrete logic and

latterly specifically tailored microprocessors, is sand-

wiched between means of coupling it to an analog

world outside. We’ll first look at the converters and then

the DSP bit.

25.18 converters

25.18.1 A/D Conversion

25.18.1.1 Resolution

A DSP processor needs a stream of digital words of

sufficient resolution to adequately portray the actual

input signal level at a given moment. This resolution is

determined by the number of binary bits in each word;

each bit corresponds to a doubling of the resolution, or

roughly 6 dB of dynamic range capability. Phone

systems typically use 8 bits (approximately 48 dB

dynamic range linear, although effectively more when

companded), the BBC’s original distribution system was

a 13 bit system (78 dB), CDs 16 bit (9 dB) and most

production and recording systems a nominal 24 bit.

The A/D conversion process is fraught, particularly

with high-resolution converters, and the actual dynamic

range is often much less than theoretically possible from

the number of bits. System noises, either from the

analog paths or crosstalk from various digital signals,

are the predominant limitation; gross errors used to

come from nonmonotonicity. Strictly speaking, a

converter should, if the input signal is increased by one

unit of the resolution, reflect this by increasing the value

of the output digital word by one bit. Often, particularly

at transitions of the major bits, this goes awry and an

Figure 25-125. Mix stage of a digital console.