Hello - noise removal help

Hi Everyone,

I’m not new to Audacity but I am new to the forums. I’ve run out of time trying to remove background noise from 2 months worth of recorded audio. A realistic goal would be for me to successfully remove the noise from 5 minutes to start. It i a microphone sitting on a desktop computer HUMMM all the time. The actual desired audio is poor to begin with (some whisper and vague talking). Maybe I set myself up for failure using this audio while following the spectrogram tutorials lol. I am at the point of shamelessly asking for someone to check it and tell me that there isn’t much that can be done. That would make me feel a lot better :slight_smile:

Thank you for Audacity, it has been a goto of mine for a long time and I am just now realizing its capabilities. Much appreciated.

  • Jeff

Is it stereo (two blue waves) or mono (one blue wave)?

Post some of it.


That technique can capture up to ten seconds and you should do that.

The actual error may have been not trying it first. Dropping straight into production recording with a new system is a very popular new user mistake. Some microphone makers set you up with unrealistic promotions.


If you want to remove the continuous (hiss/hum) noise when no-one is speaking
the the tool for that job is a “noise gate”.
You can get a noise-gate plug-in for audacity here …
https ://wiki.audacityteam.org/wiki/Nyquist_Effect_Plug-ins#Noise_Gate

NB: noise gate has no effect on the noise which occurs when the person is speaking.

Audcaity’s native noise reduction can reduce hum.
Unlike other effects, noise-reduction is a two-step process, see … https://youtu.be/PzZplo0oApI?t=52

So there is no plain, ordinary noise removal. Noise Reduction used to be called Noise Removal, but it was change because too many people expected it to actually remove noise—to zero.

Noise Reduction can gracefully reduce background noises both between and during words, but it has to take the words apart to do it. If you use too much reduction, the show will start to sound gargley, bubbly and honky.

As above, Noise Gate can do a wonderful job of noise between words, but not during them. This can give you hissy or buzzy words on a clean background, for example.

Notch Filters can be used if you have one particular tone that competes with the show. The problem comes when you have a musical tone as part of the show that falls in the notch…and vanishes.

A cousin to that is USB microphone noise (Yeti Curse/Frying Mosquitoes) That one has multiple tones that interfere with the show. We designed a special filter to get rid of it, but it, too, takes musical tones away from the performance.

So these are all patch jobs and you’re much further ahead making a good recording in the first place.


Thank you guys very much for the advice and links. I am going to keep at it for the sake of learning, at least until the effort starts to out weigh the goal lol.

I have been in spectrogram most of the night and day, the desired audio is unfortunately barely 1-2db louder than the hums; so even cleaned up the source just wasn’t quite good enough I think. BUT I at least consider myself a novice at denoising!

Here is a sample, this is one audio file that is pretty clear of the Rated-R content lol WARNING MAY OR MAY NOT BE ADULT CONTENT


Thanks Again,


Hello again!

Right now when I identify a simple line (hum or hiss) within a range, I am selecting the entire line carefully from Spectrogram view then expanding selection from beginning to end of the clip, then I hack em out with the multi edit tool. Sometimes I can get lucky (or unlucky?) and have 3-5 recurring range lines I need to do for each clip. Is there a way somehow to script these ranges so I can easily apply it to any or all clips?

Thanks in advance,