Morse Code Decoding from its spectrum

I have a Morse code signal separated from a bunch of other audio sources by Blind source separation. Now I have the frequency domain signal. Could someone suggest how to decode the Morse code from its spectrum or suggest some means by which I could obtain a time-domain representation of the same from which I may manually decode the Morse code?

n = 1000;
wn = [980 1020]/(fs/2);
[b,a] = fir1(n,wn,'bandpass')
% fvtool(b,a);
f = filtfilt(b,a,y);
afil = fft(f);
% subplot(4,1,1); plot(real(bfil)); title('Frequency Response of Input Signal');
% subplot(4,1,2); plot(real(afil)); title('Frequency Response of Filtered Signal');
% subplot(4,1,3); plot(y); title('Original Sound Signal');
% subplot(4,1,4); plot(f); title('Filtered Sound Signal');

[Daniel kiracofe]

Would not suggest trying to decode morse code by looking at it's spectrum (i.e. a single FFT). Is it physically possible? Maybe, but it will be insanely hard. Just use the time domain signal. Note, you might have some luck using a short-time fourier transform (i.e. spectrogram or related methods), but a single fft of the whole signal is not the right approach.

"some means by which I could obtain a time-domain representation of the same"

If I understand correctly, here you are just asking how to get the time domain representation of the variable 'afil'? You already have that. it's just the variable 'f'. That's what you should focus on to decode the morse code. Or maybe I misunderstood what you are asking.

Answer 1

[Arnav Mendiratta]

Daniel's answer sums it up. Using Short Time Fourier Transform makes much more sense as Daniel mentioned in his answer. If you want to learn about the algorithmic implementation of this particular use-case, this link has some good information.

If you want some reference code this file exchange submission from the community is available which converts the WAV file containing the Morse code to text.

[Krishnagopal A]

Thanks guys. I had difficulty in identifying the code with this "time-domain" representation, but by making a few changes, I was able to decode the Morse code. But, it was very difficult and demanded a lot of manual labor, as Daniel had mentioned.

I will try making use of STFT for this. Thanks again, guys!