I'm used to filtering using FIR filters with a cutoff and dB attenuations etc.

The issue I have is that I have pulses in my signal which contain similar frequencies as the noise. I'd like to totally remove the noise whilst keeping the pulses intact (no attenuation).

What I'd like to do is have a filter which is applied if the frequency content within the window is below a given magnitude and not if it's above (or really an exponential weighting depending on the magnitude of the frequency present in a given window.)

I can probably make this happen in a fairly hacky way - but is there already a developed area of DSP / mathematics - which deals with this issue? - I rather not reinvent the wheel.

It would be nice if this was fast and I didn't have to FFT each window to check the frequency / magnitude.

  • $\begingroup$ Some representative plots would help a lot with this question. Is it possible to add them please? $\endgroup$
    – A_A
    Sep 25 '20 at 15:06
  • $\begingroup$ Do you have the time knowledge of the pulses? (Rate, duration) $\endgroup$ Sep 25 '20 at 21:19
  • $\begingroup$ Nope. The idea would be - fft a section of the signal, based on magnitude of frequency content in a given range, apply filter or not. Shift window by one sample, repeat. $\endgroup$ Sep 29 '20 at 9:04

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.