I edited an audio track where I silenced out unwanted interjections from the off. However, since the original track has a low but noticeable noise the perfect silence introduced with the silence function of my audio editing tool is disturbing. What is the quickest possibility to (i) assess the level of background noise, and (ii) to replace all perfect silences with artificial noise matching the track’s normal one as close as possible? I’m using either Adobe Audition or Audacity.

  • I've toyed with a similar problem, adding noise to digitally created audio to simulate background noise/room ambiance. It involved a noise source, lowpass filter, mild compression (with the main audio fed to the sidechain) and very subtle reverb. The noise which you removed from the track is (presumably) the result of the recording environment and hardware, and would be hard to replicate accurately. As far as measuring it is concerned, you could get the decibel level during silent parts, and use that to determine the appropriate volume for your simulated ambiance, then process it to taste.
    – RICK
    Commented Jan 2, 2015 at 22:35

4 Answers 4


What is best to do if you record something specific is also record the room 'silent', so than you get the environment sound of the subject your recording. Add these layers together and you'll get your 'white noise' in a less artificial way.


I'd be more inclined to remove the white noise from the speech than try add it back in the silence.

There are some very good paid plugins that can do it - personally I use a lot of the Waves plugs, X-Noise etc - but there are freeware alternatives.

Noise Reduction is one that Google turned up. Untested but for free you can try it & see if it's any good for you.


If the original track contains a section of just the noise, you can make a loop with that and keep it playing in your new track when you mute the original one.

If not, you'll have to do some filtering to extract a noise loop.

Personally I don't think it's a good idea to generate white noise from scratch to mimic the original. It sounds different.


The whole idea with noise is that it is random sound. You can't really easily produce a noise that replicates what was there, so your best bet is to actually use the noise that was present or apply noise reduction to reduce the noise level (if you can get satisfactory results).

As Yadli mentioned, if you have a quiet part, you can sample the noise from that and chain it out. Add a constant power cross fade between the beginning of one and the end of the other so that there is no obvious jump in the noise. You may also want to very the length to avoid any noticeable beat patterns emerging if there was any inconsistency in the noise.

If you are able to reduce the noise level on the other parts though, this is preferable as keeping your noise floor low is preferable to artificially raising it for part of your audio.

Your Answer

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

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