It sounds like this“Fourier analysis only applies to waves that repeat forever, like a held note.”
Thinking it only works for repeating waves
Why it feels right
The first examples are all repeating: a sine, a square wave, a held note. The textbook formula adds up sines at multiples of a base frequency, which only makes sense if the pattern repeats at that base period. It looks as if the method is built for periodic things.
Why it’s wrong
Fourier's original series is for repeating waves, but the Fourier transform handles any signal, including a single clap, a spoken word or a photograph. Instead of a few separate frequencies it gives a continuous spread of them.
A worked example
A repeating wave with period T has a spectrum of separate lines, at 1/T, 2/T, 3/T and so on. For a signal that never repeats, such as one clap, you can pretend the pattern repeats with an infinitely long period. The spacing between the lines shrinks to nothing and they merge into a smooth curve: a continuous spectrum. That is the Fourier transform.
So a single clap has a spectrum that is broad and smooth, spread across many frequencies, as a sharp event should be. A pure, long note has a spectrum with one narrow peak. A spoken word has a spectrum that changes through the word, which is what a spectrogram plots.
The same maths works for images, where the 'frequencies' are how quickly brightness changes across the picture, and the idea sits behind JPEG. It works for any signal you can measure and write down.
How to catch yourself
- You say Fourier analysis is for music and held notes only.
- You are unsure what a single click's spectrum would be.
- You think the method needs a repeating pattern.
What to hold instead
The Fourier transform works for any signal, repeating or not. A repeating wave has separate lines; a one-off event has a continuous spectrum. The same maths handles sound, images and radio.
Where you’ll meet it
- Looking at a spectrogram of speech.
- Reading about JPEG compression.
- Wondering what a clap looks like in frequencies.
Questions people ask
- Does the Fourier transform work on non-repeating signals?
- Yes. The Fourier transform gives a continuous spread of frequencies for a signal that doesn't repeat, such as a single pulse. The Fourier series is the version for repeating signals.
- What is a spectrogram?
- A picture that shows how the frequency content of a sound changes over time. It is made by taking the Fourier transform of short, overlapping slices of the signal.