Handbook for Sound Engineers

DSP Technology 1161

31.1 Introduction

Over the past forty years, the field of digital signal pro-

cessing (DSP) has grown from its origins as a collection

of techniques for simulating the behavior of analog sys-

tems on digital computers into one of the most widely

studied and universally used tools in modern technol-

ogy. The use of DSP algorithms and implementations

has become the rule rather than the exception, with

applications in many areas such as music, communica-

tions, radar, sonar, image processing, robotics, seismol-

ogy, meteorology, and applied physics. The remarkable

growth of this discipline is largely due to two factors.

First, DSP is a powerful problem-solving tool because it

exploits the theoretical insights of discrete system the-

ory to describe, analyze, and implement many interest-

ing linear and nonlinear algorithms. Second, and more

important, there is a special relationship between VLSI

technology and DSP applications. The rapid develop-

ment of digital integrated circuit technology has contin-

ually reduced the cost and increased the speed of the

arithmetic operations necessary for DSP applications. In

addition, DSP algorithms, which have demanding com-

putational requirements but usually a very regular struc-

ture, are very well matched to the capabilities of VLSI.

Integrated circuits are making complex DSP applica-

tions possible, and DSP applications have become a

major motivating factor for building fast, complex inte-

grated circuits. Perhaps the most visible embodiments

of this phenomenon are the families of DSP micropro-

cessors commonly called DSP chips. These chips have

already had an immense impact on technology and are

currently in the process of revolutionizing much of our

industrial and technological base.

This chapter will introduce some of the important

aspects of DSP technology including the fundamentals

of DSP, the sampling process for converting analog

signals to digital signals, the algorithm development

process, and an introduction to programmable DSP

devices. References are provided for finding additional

information.

31.2 Digital Signal Processing

DSP is a technology and technique for analyzing and
extracting information from signals, synthesizing sig-
nals, and manipulating signals. The acronym DSP is
often used as both a noun and an adjective. DSP also
often stands for digital signal processor—the actual
microprocessor/computer that is used to implement the
system. Common applications of DSP include cellular
telephones, MP3 players, surround sound receivers,

compact disc players, digital cameras, answering

machines, and modems.

As with many disciplines, there are different

perspectives and different layers of abstraction from

which to explore DSP. For the purposes of this chapter,

DSP will be approached and introduced from the theo-

retical, physical, and embedded software perspectives.

The theoretical perspective is concerned with the

question “is something possible” and is built from

fundamentals of DSP theory. This foundation includes

linear system theory, complex number theory, and

applied mathematics. The theoretical level provides a

common language for DSP researchers to study and

advance the state of the art.

The physical perspective is concerned with the

devices that are used to implement DSP systems. These

devices include the programmable digital signal proces-

sors that perform mathematical operations at a very high

speed, and the details of converting an analog signal

into a digital signal and then back to an analog signal.

The embedded software perspective is concerned

with the actual software that makes the digital signal

processors perform the desired tasks. This software is

called embedded because it is executed internally on the

DSP device and is only user accessible through some

user interface, effectively hidden or embedded in the

product, hiding the implementation details from the user.

31.3 DSP Signals and Systems Theory

The concepts of signals and systems are critical to an

understanding of DSP. Signals can be a function of con-

tinuous time (i.e., analog) or of discrete time. Continu-

ous-time signals have a signal value at any given instant

of time while discrete-time signals only have a signal

value at discrete instants of time. Values of dis-

crete-time signals between the samples are determined

by mathematically interpolating between the known

sample values.

Signals represent the data that is to be processed.

Examples include an audio file that needs to be

compressed for low bit-rate storage or transmission or

an image that will be searched for a particular object. A

system is a transformation that maps an input signal (or

multiple input signals) to an output signal (or multiple

output signals)—i.e., the black box that maps inputs to

outputs. In the music compression example, the output

signal could be an MP3 file that was created by

compressing an input signal. In the image example the

output signal could simply be a yes/no decision along

with positioning information. DSP systems are typically

designed from simpler subsystems much like computer