I have signal which contains some noise only regions (glitch with the recording it seems).

The question is:

  1. I would like to identify the discrete time samples where this noise part starts and ends. There are about 3 to 4 seconds of this noisy part available, there are 4 to 5 such instances, the total recording length is 153 seconds. The sampling frequency is 48 kHz.
  2. I would also like to get to characterize the spectra of noise so that I can possibly try and remove it. How can I do this?
  • $\begingroup$ We'll need to know at least a bit about your noise! $\endgroup$ – Marcus Müller Apr 11 '20 at 10:18
  • 2
    $\begingroup$ @Dsp guy sam, I've answered a similar question here dsp.stackexchange.com/questions/66177/… so you can have a look. $\endgroup$ – dsp_user Apr 11 '20 at 10:21
  • $\begingroup$ @dsp user that is close to what I am looking for, but how do I find the start of the noisy portion in discrete domain $\endgroup$ – Dsp guy sam Apr 11 '20 at 17:34
  • $\begingroup$ @Dsp guy sam, can you open your file in a sound editor and start from there (i.e. label the noisy parts manually)? If you're looking for a fully automatic algorithm, then you'll need to know more about the noise. $\endgroup$ – dsp_user Apr 11 '20 at 19:24
  • $\begingroup$ @dsp user, what should I be knowing about the noise to automatically find the start of this noise phase $\endgroup$ – Dsp guy sam Apr 11 '20 at 20:28