About time-frequency analysis

I am having trouble understanding how the time width of the Kaiser window is determined from each parameter used in "pspectrum" for time-frequency analysis. Currently, I am using the code below:

pspectrum(W_mix,Fs,'spectrogram','FrequencyLimits',[0 1.2e6],

'FrequencyResolution',9e4,'Overlap',99,'Leakage',0.75,'MinThreshold',-95);

Here, w_mix is a waveform data of size 1×15000, and Fs is the sampling frequency, which is 1×10^8. I understand that the shape factor β\betaβ is 10 in this case, but how is the time width of the window function determined? According to the previous research I am referring to, the time width is supposed to be 20.52 μs, but I do not understand the calculation method. Could you please explain it to me?

Answer 1

[arushi]

Hi Haru,

To understand how the time width of the Kaiser window is determined in the context of the pspectrum function in MATLAB, we need to look at several factors, including the sampling frequency, the desired frequency resolution, and the properties of the Kaiser window.However, you mentioned that the expected time width is supposed to be 20.52 μs. This discrepancy suggests that additional context or parameters might be influencing the calculation, such as the way the pspectrum function internally adjusts the window length or the effect of the Leakage parameter.The Leakage parameter affects the window's mainlobe width and can influence the effective time width. A higher leakage value typically results in a narrower mainlobe (higher frequency resolution) but with higher sidelobe levels.

Hope this helps.