Chapter 9Practical Applications

You've built the machinery. Here's where it actually runs — you interact with the Fourier transform dozens of times a day without noticing.

Try it: tune a radio

A radio receiver doesn't "hear" one station and ignore the rest — every station's signal arrives at the antenna at once, all mixed together, exactly like Chapter 2's sum of waves. Tuning the dial is choosing which frequency band to isolate and listen to.

Turn the knob to sweep across the dial. The nearest station locks in and plays its tone below.

Other stations (ignored) Tuned-in station What your speaker plays

The full loop, once more

A signal is sampled respecting the Nyquist limit (Ch. 6); the FFT turns those samples into a spectrum (Ch. 7) using the same rotating-phasor idea as the epicycles (Ch. 4); that spectrum — of a sound (Ch. 5) or an image (Ch. 8) — is edited, filtered, or compressed directly; and an inverse transform, the same formula run backwards, rebuilds a time- or space-domain signal as a sum of sine waves (Ch. 3). Every example on this page is that loop, wearing a different costume.

Where to go from here: this site stops at the DFT/FFT and its 2D extension, but the same ideas keep going — the convolution theorem (filtering in time ≡ multiplying in frequency), windowing and spectral leakage (what happens when a signal doesn't divide evenly into your sample window), the continuous Fourier transform for non-repeating signals, and spectrograms (a Fourier transform repeated over sliding time windows, the basis of sound visualizers and speech recognition) are the natural next steps.