I'm asking myself what is the best way to low pass a signal which only has a few hundred samples? Or is there a proper way at all?

It must not be a sound signal but I used Audacity to display the samples. And I applied their low pass effekt (2.img). Problem here, the signal was shiftet to the right. Is their any way to implent an low pass filter without shifting the samples? Thanks in advance. raw Signal

low passed

Now a bit more detailed.
First of all I start with a spectrogram(3.Img) where I try to follow one of those maxima. That gives me the blue line. But my algorithm is not perfect, thats why I dont get a very smooth blue line. And I need this blue line to be a smooth curve. Therefore I thought, let's pretend that this blue line is a sound signal and try to smooth it through lowpass filtering.
What do I mean with short:
In this example it is less than 200 samples but it depends on the soundfiles length.

What do I mean with low:
I need this blue line to be smooth. And since I thought it would be possible to pretend like it is a sound signal I could use lowpass filtering. But I cant tell u a discrete value and I cant say anything about samplerate, since it is no real sound signal.

I hope it is more clear now. And thanks for your answers so far. spectrogram

  • 1
    $\begingroup$ Any causal filtering has a delay as effect – but there's no need for your filtering to be causal if it's done "offline", i.e. on recorded data. So, just shift by the group delay of your filter (assuming that is constant). The question really is what "short" signal means to you (in samples) and what "low" in low pass signal means to you (in fraction of sampling rate). Without that, no assessment on what is possible or not can be done. $\endgroup$ – Marcus Müller Mar 2 '19 at 0:12
  • $\begingroup$ Agreeing with Marcus. $\endgroup$ – Digiproc Mar 2 '19 at 12:15
  • $\begingroup$ Thanks for your answers so far. I tried to explain the context a bit more. $\endgroup$ – cube Mar 2 '19 at 15:14

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.