Mathematical algorithm for rapid frequency decomposition of audio signals — foundation for EQ, spectrum analyzers, real-time processing. Makes invisible frequencies visible on screen.
You're at the editing suite, sitting in front of the monitor, wanting to know exactly where that annoying hum in the dialogue recording is coming from — and then you open the spectrum analyzer. What you see is real-time FFT: the mathematical decomposition of the audio signal into its individual frequency components. The Fast Fourier Transform makes the invisible visible — and does it lightning fast.
The FFT works on a simple principle: every audio signal, no matter how complex, can be represented as a sum of sine waves of different frequencies. Instead of analyzing the whole signal like a mush, the FFT breaks it down into its constituent parts. A 60 Hz hum, 2 kHz hiss, 200 Hz rumble — everything is made visible individually. Without FFT, real-time EQing, spectrum analysis, and the entire modern audio workflow wouldn't function. Your DAW, your audio player with a visualizer, any plugin interface that displays frequencies — they all use FFT.
On set, you're less interested in the theory, but much more in the application. The sound mixer has recorded a busy street — wind, traffic, background noise. In the edit, the spectrum analyzer shows you exactly where the problems lie. You look at the FFT graph, see the frequency peak at 120 Hz (typical mains hum in Europe), and set a narrow EQ notch precisely there. Effective, surgical. Without FFT, you'd be blindly twiddling EQ knobs, hoping it gets better.
In practice, you should know: FFT resolution is a trade-off. The longer the analysis window, the higher the frequency resolution — but the lower the temporal resolution. If you want to see *when* exactly a noise peak occurs, you need to make the windows smaller. This is relevant for live audio processing, such as noise reduction in real-time processes. The FFT algorithm itself — named after Cooley and Tukey — dramatically reduced the computation time for frequency analysis. This made real-time audio possible. If the raw Fourier transform had to be used, modern live streams and online calls wouldn't work yet.
Don't use FFT as theoretical ballast. Use it as a tool: open the spectrum analyzer, locate the problem, intervene precisely. Combined with waveform editing (see there), FFT-based audio analysis is one of your strongest allies for clean, professional sound.