1

I'm trying to figure out a way to manage background noise in audio without having to resort to noise reduction. I'm using audacity to post-process spoken audio. The levels are pretty low, with peaks never breaking -20db.

The background hiss doesn't even register in the unprocessed raw audio, but if I normalize or run the compressor with normalization afterwards, the hiss gets amplified to around -43db. It seems the noise floor is ignored during normalization.

Is there a way to normalize audio with a noise floor, so I can leave anything below -55db or so untouched?

  • 2
    This could sort of be summarised thus: "how do I reduce my noise-floor without actually doing noise-reduction"? Unfortunately there is no answer to this. You actually need to use a noise-reduction plugin, or go back to the recording stage and use a technique which gives you a better signal-to-noise ratio. This can be achieved by using a more directional mic technique or simply getting the mic closer to the signal source. – Mark Jan 22 at 9:05
1

The trouble with Normalisation is it is undiscerning, it simply increases all gain until one peak reaches 0dBFS - & brings the entire noise-floor with it by the same amount.

Adding compression is only going to make this worse, as it's limiting the peaks, allowing the noise floor to come up further.

The simplest method is going to be to use the 'opposite' of a compressor - that is an Expander. A noise gate is one type of expander, but there are ones that not only open up the 'gate' but then also amplify the remaining signal. Wikipedia covers both in Dynamic Range Compression

An expander will ignore signals below threshold, but then increase the gain once above threshold - thereby leaving your noise floor where it is, yet increasing where the speech occurs.

The down-side, unless you actually employ a true noise-reduction plugin, is that in these areas, the noise floor will also increase. It will only stay silent in the periods determined by the threshold as 'silent'.

-3

Remove the noise first. Then compress over your threshold and add make up gain.

Depending on the noise and level you may not be able to achieve what you want without a better source recording.

Normalization raises everything so the peaks are at -1 in audacity. You can choose another peak if you don't like the default.

You could try to remove noise a second time before compressing and adding make up gain. Not sure you could achieve much with that but worth a try once.

  • Just a comment on your style of answering. It's not really appropriate to be asking people what their "hangups" are, or what they are afraid of. This is an ad-hominem comment and not really what stack exchange is all about. Try and limit your answers to on-topic information and you will avoid being down-voted. – Mark Jan 25 at 15:12
  • Sometimes. I think that was appropriate in the answer based on the question. So NOT adhominem at all just necessary to fully understand why they are having problems to better offer a solution. The bullies and cowards will downvote me just because I exist and they because they can which makes them feel good about themselves. – edwina oliver Jan 25 at 16:48
  • Then try and word your responses so that they are less confrontational.You will find that it's not "bullies and cowards" downvoting you. Downvoting is not "bullying", it is a community response to the quality of your answers. Improve the answers and it will likely improve your reputation. That's how this site works. You'll get it eventually. – Mark Jan 26 at 1:07
  • @edwina - Mark's comment is correct. We do not tolerate rudeness here. It is not bullies downvoting you - they are helping prevent bullying. – Rory Alsop Jan 29 at 20:43

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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