Connexions

You are here: Home » Content » Short Time Fourier Transform
Content Actions

Short Time Fourier Transform

Module by: Ivan Selesnick

Summary: Introduction to the Short Time Fourier Transform, which includes it's definition and methods for its use.

Short Time Fourier Transform

The Fourier transforms (FT, DTFT, DFT, etc.) do not clearly indicate how the frequency content of a signal changes over time.
That information is hidden in the phase - it is not revealed by the plot of the magnitude of the spectrum.
Note: To see how the frequency content of a signal changes over time, we can cut the signal into blocks and compute the spectrum of each block.
To improve the result,
  1. blocks are overlapping
  2. each block is multiplied by a window that is tapered at its endpoints.
Several parameters must be chosen:
  • Block length, RR.
  • The type of window.
  • Amount of overlap between blocks. (Figure 1)
  • Amount of zero padding, if any.
STFT: Overlap Parameter
stftfigs.png
Figure 1
The short-time Fourier transform is defined as
Xωm=STFTxn≔DTFTxn-mwn=∑n=-∞∞xn-mwnⅇ-ⅈωn=∑n=0R-1xn-mwnⅇ-ⅈωn X ω m ≔ STFT x n DTFT x n m w n n x n m w n ω n n 0 R 1 x n m w n ω n (1)
where wn w n is the window function of length RR.
  1. The STFT of a signal xn x n is a function of two variables: time and frequency.
  2. The block length is determined by the support of the window function wn w n .
  3. A graphical display of the magnitude of the STFT, |Xωm| X ω m , is called the spectrogram of the signal. It is often used in speech processing.
  4. The STFT of a signal is invertible.
  5. One can choose the block length. A long block length will provide higher frequency resolution (because the main-lobe of the window function will be narrow). A short block length will provide higher time resolution because less averaging across samples is performed for each STFT value.
  6. A narrow-band spectrogram is one computed using a relatively long block length RR, (long window function).
  7. A wide-band spectrogram is one computed using a relatively short block length RR, (short window function).

Sampled STFT

To numerically evaluate the STFT, we sample the frequency axis ωω in NN equally spaced samples from ω=0 ω 0 to ω=2π ω 2 .
∀k,0≤k≤N-1: ω k =2πNk k 0 k N 1 ω k 2 N k (2)
We then have the discrete STFT,
X d km≔X2πNkm=∑n=0R-1xn-mwnⅇ-ⅈωn=∑n=0R-1xn-mwn W N -kn= DFT N xn-mwn|n=0R-10,…0 ≔ X d k m X 2 N k m n 0 R 1 x n m w n ω n n 0 R 1 x n m w n W N k n DFT N n 0 R 1 x n m w n 0,…0 (3)
where 0,…00,…0 is N-RNR .
In this definition, the overlap between adjacent blocks is R-1 R 1 . The signal is shifted along the window one sample at a time. That generates more points than is usually needed, so we also sample the STFT along the time direction. That means we usually evaluate X d kLm X d k L m where LL is the time-skip. The relation between the time-skip, the number of overlapping samples, and the block length is Overlap=R-L Overlap R L
Problem 1
Match each signal to its spectrogram in Figure 2.
sgrams1.png
Subfigure 2.1
sgrams2.png
Subfigure 2.2
Figure 2
[ Click for Solution 1 ]
Solution 1
[ Hide Solution 1 ]

Spectrogram Example

stft_x.png
Figure 3
stft_256.png
Figure 4
The matlab program for producing the figures above (Figure 3 and Figure 4). 
	  

	  % LOAD DATA
	  load mtlb;
	  x = mtlb;

	  figure(1), clf
	  plot(0:4000,x)
	  xlabel('n')
	  ylabel('x(n)')

	  % SET PARAMETERS
	  R = 256;               % R: block length
	  window = hamming(R);   % window function of length R
	  N = 512;               % N: frequency discretization
	  L = 35;                % L: time lapse between blocks
	  fs = 7418;             % fs: sampling frequency
	  overlap = R - L;

	  % COMPUTE SPECTROGRAM
	  [B,f,t] = specgram(x,N,fs,window,overlap);

	  % MAKE PLOT
	  figure(2), clf
	  imagesc(t,f,log10(abs(B)));
	  colormap('jet')
	  axis xy 
	  xlabel('time')
	  ylabel('frequency')
	  title('SPECTROGRAM, R = 256')

Effect of window length R

Narrow-band spectrogram: better frequency resolution
stft_512.png
Figure 5
Wide-band spectrogram: better time resolution
stft_128.png
Figure 6
Here is another example to illustrate the frequency/time resolution trade-off (See figures - Figure 5, Figure 6, and Figure 7).
Effect of Window Length R
sgramsR_1.png
Subfigure 7.1
sgramsR_2.png
Subfigure 7.2
Figure 7

Effect of L and N

A spectrogram is computed with different parameters: L∈110 L 1 10 N∈32256 N 32 256
  • LL = time lapse between blocks.
  • NN = FFT length (Each block is zero-padded to length NN.)
In each case, the block length is 30 samples.
Problem 2
For each of the four spectrograms in Figure 8 can you tell what LL and NN are?
sgramsLN_1.png
Subfigure 8.1
sgramsLN_2.png
Subfigure 8.2
Figure 8
[ Click for Solution 2 ]
Solution 2
[ Hide Solution 2 ]
LL and NN do not effect the time resolution or the frequency resolution. They only affect the 'pixelation'.

Effect of R and L

Shown below are four spectrograms of the same signal. Each spectrogram is computed using a different set of parameters. R∈1202561024 R 120 256 1024 L∈35250 L 35 250 where
  • RR = block length
  • LL = time lapse between blocks.
Problem 3
For each of the four spectrograms in Figure 9, match the above values of LL and RR.
stft.png
Figure 9
[ Click for Solution 3 ]
Solution 3
[ Hide Solution 3 ]
If you like, you may listen to this signal with the soundsc command; the data is in the file: stft_data.m. Here is a figure of the signal.
stft_signal.png
Figure 10

Comments, questions, feedback, criticisms?

Send feedback