2

I have read the following sentence in a documents talk about the drawback of short time Fourier transform, and he is said :

the drawback is that once you choose a particular size for the time window, that window is the same for all frequencies

So what is the relation between frequencies and the size of window. If we have a high frequency component in a part of a signal how will not be able to detect this frequency if the size of the window is not smaller/bigger enough?

Furthermore, he is said about wavelet transform :

Wavelet analysis allows the use of long time intervals where we want more precise low-frequency information, and shorter regions where we want high-frequency information

I feel that the answer has a relation with nyquest rate somehow

1 Answers1

1

For sampled data, the number of orthogonal sinusoidal FT basis vectors below half the sample rate increases with the length of the STFT window, and the bandwidth of each DFT/FFT result bin for each basis vector decreases. If the window is too short, then each DFT result might detect not only your high frequency component of interest, but a greater bandwidth of adjacent frequencies.

hotpaw2
  • 70,107
  • 14
  • 90
  • 153