An FFT analyzes a chunk of audio as if it looped forever, and a chunk cut mid-wave loops with a click. That click smears energy across the spectrum, called leakage. Window functions fade the chunk’s edges to soften the cut: Hann, Hamming, Blackman, each a different tradeoff between sharp frequency peaks and low smear.
The window also sets an FFT’s time-versus-frequency personality alongside its size: long windows resolve close pitches but blur onsets, short ones catch the hit but blur the harmony.
Every spectrum you have ever seen was shaped by a window choice, usually a Hann, usually unmentioned. Analyzer displays that look crisp are as much window engineering as data.