The name given to windows used to analyse/synthesize a signal with wavelet transforms. Used mostly in the analysis of non-stationary signals whose frequency content varies with time. The variable length windows with which the signals are analysed are called wavelets.
Most signals in nature are non-stationary, i.e, they have a time-varying frequency content. This sort of signal is largely useless to an analysis with FFT because of the nature of its frequency content.
An FFT gives 100% resolution in the frequency domain (i.e, you know what frequencies the signal is made of), but 0% resolution in the time domain (you have no information on when that frequency component occurred). STFT (Short Time Fourier Transform) seeks to provide resolution in time and frequency domains simultaneously by analysing the signal in fixed length windows. This provides equal resolutions in both domains, not the best of solutions.
Wavelet transforms analyse the signal with different length windows (wavelets with different resolutions) which gives a multi-dimensional representation of the signal. There are many different wavelet base functions, but the more popular ones are Daubechies and Haar.
