How can i generate spectrogram for a complex signal?

I am having a complex signal as recived signal at receiver. e.g. (X_noise_signal=0.1+0.3i, 0.2-0.4i,................). I want to generate a spectrogram for this signal so that i can use it at input layer of a Convolutional Nural Network architecture. Kindly provide your suggestions.

regards,

Chirag Gupta

Answer 1

[Chunru]

Use spectragrom which works both for real and complex data:

a = randn(8192, 1) + 1i*randn(8192, 1) + cos(2*pi*0.25*(0:8191)');      % complex data
spectrogram(a, 512, 256, 512)

[Chirag Gupta]

Thanks for your reply. But my signal dont have cos components its a (3932160 X 1) complex vector. This command is not working for me. It will also help if i can represent it into a matrix form or a 2 channel image (one channel as real part and other channel as imagenery part)

[Chunru]

You replace "a" with your data. The data here is for illustration only. For two-channel signal, you make the choice of having two spectrogram individually, or findong the average spectrogram, or mixing the multiple channel signals and then computing spectragram. Decision is yours.