How can I filter out noise from a signal from representative noise samples?

Hi there!

I am looking to denoise the frequency spectrum of a noisy signal. Because of the nature of the data, I have two sets: one where the only thing measured is assumed to be the background noise (there's actually a few representative ones), and another where I have a signal recorded over that noise.

I was wondering if anyone knew any good ways to remove that background noise from the signal given that I have a sample of it. I have already tried filtering in just the frequency domain, but the signal is still noisy around the frequencies I am trying to keep.

Answer 1

[Abderrahim. B]

Hi!

Since you have the noise known and the measured signal to filter, I suggest to use adaptive filtering. DSP System Toolbox comes with different types of these filters such as LMS, RLS, NLMS, ....

Please refer to these examples from the toolbox documentation to learn how to use adaptive filters.

Hope this helps.

[Michael Vaccaro]

Thank you! I'll look into it :)