-3

Just curious, if my environmental noise is very stable and doesn't change much, and I'm collecting sound data in this environment. Can I collect the environment noise first, and then collect the sound data (with the environmental noise of course), after that, using some methods to remove the environmental noise from the sound collection data?

Is there any way I can perform such process by python? Or is there any similar algorithms?

After a few months, I found my question is how to do the sound separation by neural network. I am glad to find the solution here. https://sigsep.github.io/open-unmix/#paper It's called open unmix.

Xudong
  • 441
  • 5
  • 16
  • This assumes that the noise doesn't change much between the two samples. If that's a valid assumption, why wouldn't this work? – Barmar Aug 29 '19 at 19:08

1 Answers1

0

You can use SVD(single value decomposition) to seperate your audio. This shows how to do it in matlab. But you want a python solution. I searched and found this. I hope it solves your problem.

EBIN JOSEPH
  • 66
  • 2
  • I just checked that post. And I finally found this problem is too broad and really hard to get a perfect solution. Anyway, thanks very much for giving such a link! Also, I'm wondering is there any progress about this problem during these five years? – Xudong Aug 30 '19 at 02:32