7

I've been recording podcasts and screen casts for a while now. I've always ran all my audio through Audacity and used the noise cancellation to clean up the audio. I've been happy with the results generally.

I was attending a conference session on recording videos and they suggested recording some silent room noise, and then mixing it in the entire recording so the background noise was consistent after editing. I asked him why he didn't just remove all the noise instead. He said that the background noise sounded more natural, and noise removal distorted the audio too much.

I know that not dealing with the slight amounts of background noise is not a good solution for edited audio since the change from noise to no noise, or different noise is very noticeable.

I don't have any training in audio production, so I am curious what is the recommended method of dealing with this. Is adding bed noise the preferred method? Does using active noise cancellation really distort the audio that much (and I just don't notice it)?

Thanks!

Improve this question
4
  • 4
    For speech, I've never found the slight distortion due to noise removal bothersome.
    – Matthew Read
    Commented Dec 20, 2010 at 20:09
  • 1
    And same with music. I compress all my music, and then apply Noise Removal. Very rarely do any side-effects happen, and that's because the original music quality is bad.
    – muntoo
    Commented Dec 20, 2010 at 23:12
  • In contrast, I would vote in favour of keeping the background sound and then mixing in appropriately. it's very important to have natural, consistent audio -- it will help viewers effectively ignore flaws in the video (amazing what poor quality of video you can tolerate when the sound is good). I HATE the aliased, digital sound of excessively noise-reduced vocals; it sounds unnatural and blends syllables together when not applied well. I always prefer a bit of location sound with all of the speech sounds audible, just blend any scene transitions. (I am an audio engineer & sound recordist)
    – Chris Woods
    Commented Sep 15, 2011 at 12:29
  • 1
    Noise doesn't bother anyone as much as changes in the noise. Hiding varying noise under added steady noise makes the changes less attention-grabbing (hopefully). Like a lamp going off suddenly in a room filled with sunlight - no one will notice.
    – DarenW
    Commented Oct 31, 2012 at 5:11

3 Answers 3

Reset to default
5

While I don't know of any standard practice for this, I'd say use your ears. You've tried it with the noise removal tool, now try it by adding room noise as best you can - if indeed the change is too noticeable, then scrap it.

If you're not sure if your ears are deceiving you or not, have some friends listen to it on both your system and theirs, before and after processing. See if they think the noise cancellation is distorting too much.

If it seems to be working for you, I can't think of any reason to change it!

Improve this answer
4

SINAD = (P_signal + P_noise + P_distortion)/(P_noise + P_distortion)

The signal-to-noise and distortion ratio might be useful to think about.
You are either adding additional noise or you are kind of converting noise to distortion...

Adding additional noise would have more effect on the SINAD ratio than converting noise into distortion. The same can be noticed with our ears, you hear a bit of noise much easier than a bit of distortion. I would rather have a clean speech that doesn't contain noise than a (non-)constant noise when using in-ears.

(Unless you are turning the volume down so the noise would be less loud and thus less disturbing...)

Improve this answer
4

Recording a bit (30 seconds to 1 minute) of ambient "silence" is common when shooting interviews, documentaries, locations shots, etc. Utilizing it goes a long way to making edits that are not jarring or distracting.

Improve this answer

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.