時間周波数解析 "pspectrum" について

時間周波数解析に使う　"pspectrum"　について、各パラメータからどのようにKaiser窓の時間幅が決まるのかわからず困っています。

現在私は下記のコードを使用しています。

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

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

＊ここで、w_mixは1×15000の波形データ、Fsはサンプリング周波数で1×10^8です。

この場合、形状係数βが10となるのはわかるのですが、窓関数の時間幅はどのように求まるのでしょうか。

参考にしている先行研究によると、時間幅は20.52μsとなるらしいのですが計算方法がわかりません。ご教授よろしくお願いいたします。

Answer 1

[Suraj Kumar]

Hi Haru,

From what I gather, you want to figure out how the time width of the Kaiser window is determined with the “pspectrum” function in MATLAB.

For this, you can go through the following steps and the relevant code snippets:

1. Create the spectrogram plot using the “pspectrum” function with the specified parameters.

W_mix = rand(1, 15000); % Example waveform data 
     Fs = 1e8;   
     FrequencyResolution = 9e4;   
     % Calculate the spectrogram and plot it 
     pspectrum(W_mix, Fs, 'spectrogram', ... 
           'FrequencyLimits', [0 1.2e6], ... 
           'FrequencyResolution', FrequencyResolution, ... 
           'Overlap', 99, ... 
           'Leakage', 0.75, ... 
           'MinThreshold', -95);

2. Retrieve the handle to the current spectrogram plot using “gca” function and extract the time vector from the plot’s children.

h = gca; 
spectrogramData = h.Children; 
timeVector = spectrogramData.XData;

3. Then determine the length of time window by taking the difference between the first and last points of time vector and dividing by the length.

% Calculate the time width of the window 
windowLength = length(timeVector); % Length of the time window 
timeWidth = (timeVector(end) - timeVector(1)) / windowLength; 

Refer to the output below for more clarity:

For more information on “gca” function and graphic objects in MATLAB, you may refer to the following documentations:

• https://www.mathworks.com/help/matlab/ref/gca.html
• https://www.mathworks.com/help/matlab/creating_plots/graphics-objects-overview.html

I hope this is helpful.