Window (signal processing)
Appearance
![]() |
In signal processing, a window is used to analyze a finite amount of data in the frequency domain. Various types of windows (and the size of the first side-lobe) are the following:
- rectangular window: -13 dB (is the same as "not using a window")
- Bartlett window -27 dB (triangular: 0 at ends, 1 at center)
- Bartlett-Hanning window -32 dB (raised cosine: 0 at ends, 1 at center)
- Blackman window -58 dB (2 cosine waves ... 0 at ends, 1 at center)
- Blackman-Harris window
- Bohman window
- Chebyshev window
- Gaussian window
- Hamming window -43 dB (raised cosine: +0.08 at ends, 1 at center)
- Hann window
- Kaiser window (... uses a Bessel function ...)
- Tukey window
The rectangular window gives the best frequency accuracy (narrowest main lobe) at the cost of the highest side-lobes. The Blackman window gives the lowest side-lobes at the cost of the widest main lobe.