What is a signal?

A signal is any quantity that varies — a voltage over time, brightness across an image, pressure in the air carrying this sentence to an ear. Signal processing is mathematics' answer to one question: what can we learn from, or do to, a thing that varies?

Build the intuition

Signals are functions wearing work clothes

You already know the math object: a signal is a function — input time (or position), output value. x(t) for a microphone voltage, brightness(x, y) for a photo. Everything you learned about functions and graphs applies verbatim; signal processing just puts those functions to work.

x(t): \text{time} \to \text{value}

Continuous vs discrete: the great divide

Nature's signals are continuous — defined at every instant, written x(t). Computers can't hold infinitely many values, so they keep snapshots at regular ticks: a discrete signal x[n], a numbered list. Round brackets: analog world. Square brackets: digital world. The bridge between them — sampling — is so important it gets two lessons of its own.

x(t) \;\text{(continuous)} \qquad x[n] \;\text{(discrete)}

The signals worth knowing by name

A few characters recur everywhere: the sinusoid A sin(2πft + φ) — pure tone, the atom of this whole subject; the step (off, then on — a switch flipping); the impulse (one instantaneous kick — the system-tester you'll meet in the convolution lesson); and noise (the random hiss that statistics taught you to tame). Most real signals are casts of these characters.

See it move

InteractiveBuild a wave

Amplitude1.5

Frequency1

y = 1.5 sin(1x): swings 1.5 high (amplitude — loudness, brightness, tide height) and repeats every 6.28 units (period — pitch, frequency, season length).

The sinusoid — the one signal to know intimately. Amplitude is its size, frequency its rhythm: every parameter of the wave you'll spend this course decomposing things into.

A worked example

Describe a heartbeat as a signal

An ECG measures heart voltage: a continuous signal x(t), roughly periodic at ~1.2 Hz for a resting adult.
The monitor samples it 360 times per second, producing the discrete x[n] = x(n/360) — square brackets, a list a computer can store.
A cardiologist's questions are signal questions: What's the repetition frequency? (heart rate.) Are there unusual spikes? (arrhythmia.) How much is noise? (electrode quality.)
One trace, three professional questions — all answerable with this course's tools.

Out in the world

Your phone is a signal-processing factory

In one voice call: the microphone turns pressure into a continuous voltage, a converter samples it 8,000+ times a second, codecs compress the list, radio circuits modulate it onto carrier waves, and the reverse chain rebuilds a voice at the far end. Every block is a chapter of this course.

Common confusion, cleared

“Digital signals are worse copies of analog ones.”

Under the right conditions (the sampling theorem, in the Fourier course), the samples contain everything — the continuous signal is perfectly recoverable. Digital isn't lossy by nature; careless digitization is.

“Signal processing is an engineering topic, not math.”

It's functions, trigonometry, calculus, complex numbers, and linear algebra — pointed at sound and data. If you've followed this platform's courses, you already own most of the toolbox.

Check yourself

PracticeQuick check

x[n] with square brackets signals…

Recap

A signal is a varying quantity — mathematically, a function.
x(t) is continuous (nature); x[n] is discrete (computers).
Sinusoids, steps, impulses, and noise are the recurring cast.

Progress saves in this browser.