2
$\begingroup$

What is the way to address the removal of white noise from a signal?

Extra Details Hi I also agree the details are as follow: The signal is a acoustic or a sound wave. The the power associated with the frequency of the noise is distributed evenly across the whole frequency range (white noise) The actual signal characteristic is not known as its a inverse problem. But the actual frequency of the signal is a part of the total frequency. Power associated with the signal is higher than the noise can be seen in a spectogram. Actuallty I also cant figure out how to extract the coefficients of the power/frequency (db/Hz) from the spectogram. Can you suggest a way? see the graphs in the link

http://www.mathworks.com/help/signal/ref/spectrogram.html?refresh=true

Actual problem is to find the time of arrival of the actual signal

Improve this question
$\endgroup$
4
  • 1
    $\begingroup$ Welcome to SE.DSP. Additional details are required on the properties of your signal, for moore efficient answers $\endgroup$
    – Laurent Duval
    Feb 25, 2016 at 9:46
  • $\begingroup$ Hi I also agree the details are as follow: $\endgroup$
    – Avik
    Feb 25, 2016 at 15:59
  • $\begingroup$ If the signal is narrowband (not a wide range of frequencies), apply a window to the FFT. That's a crude start; using established filter techniques will be better. $\endgroup$
    – Carl Witthoft
    Feb 26, 2016 at 12:22
  • $\begingroup$ Your question has beeen answered. Do not hesitate to vote for the useful ones and accept the most suitable $\endgroup$
    – Laurent Duval
    Feb 9, 2017 at 17:12

1 Answer 1

Reset to default
1
$\begingroup$

Without further details on the signal and the type of processing you want to apply, the two most basic approaches are:

If your signal is non-stationary, a time-frequency (spectrogram) or time-scale (wavelet) decompositions might help. In their domain, signal and noise are often better separated, and you can there apply Wiener filering, or alternative forms of thresholding, or more involved source separation techniques.

Wiener filtering in the wavelet domain are generated a lot of works, and codes, that you can test.

Improve this answer
$\endgroup$
2
  • $\begingroup$ Thanks I have added more details to the question Please Help $\endgroup$
    – Avik
    Feb 25, 2016 at 16:11
  • $\begingroup$ I recommend that you provide a link to a tech description of Wiener filter so the OP can learn how it works, not just that it (might) work. $\endgroup$
    – Carl Witthoft
    Feb 26, 2016 at 14:56

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.