Remove a specific sound

Hello, I am wondering if it is possible to remove a very specific sound from a recording. The recording is of a skype call, and the application is used to record it captures all audio on that plays on the computer. I accidentally played some other sounds while the call was recording, so there are some very specific background noises in the audio. What I would like to do, is use a sample of the original sound effect that was played by mistake, and remove it from the audio of the recorded conversation, would this be possible?

What I would like to do, is use a sample of the original sound effect that was played by mistake, and remove it from the audio of the recorded conversation, would this be possible?

:frowning: Probably not.

Under certain precise conditions you can subtract a sound from a track. For example, if you record yourself saying “Hello” and you mix that file with a music file, you can take that original-unmixed “Hello”, use the invert effect, and mix that inverted “Hello” with the previously-mixed file and the two “Hellos” will cancel perfectly. But, it has to be an exact inverted copy with no level changes and no timing or phase changes, or clipping in the mixed file, etc. (In a real world situation like this, you’d probably also have the original unmixed music file so there would no need to do it.)

Here’s what won’t work… Record yourself saying “Hello”. Record yourself saying “Hello” again and subtract that from the original (with the invert & mix method). This time, the two sounds will NOT cancel. In fact, in this case subtraction will sound exactly like addition (regular mixing) and you’ll hear yourself and your imaginary twin saying “Hello” together. …This is one of those cases where "the difference in the sound is nothing like “the sound of the difference”.

Audacity’s Noise Reduction feature can do this. I’ve used it to remove the hissssss of a modem that marred a telephone interview I once recorded.

1). Find the specific sound in the WAV file you want to remove. Highlight the one that’s most isolated from the sound you actually want to keep.

2). Go into effects and pull up Noise Reduction.

3). Click on “Get Noise Profile.” This “samples” your sound.

4). Now “select all” on your track and go back into Noise Reduction.

5). Set it at around 10 (top level), 3 (middle), and 2(bottom). Click “OK.”

The Noise Reduction will then kick in and it should remove at least some of what you want gone, if not all.

If it doesn’t work as well as you need, try a different sample of the audio you want removed (Step 1) or tweak the settings, specifically raising the number of that top level in Noise Reduction.

After much discussion and testing we changed the default setting for the upcoming 2.1.2 release to 12,6,3 - you may want to try that. That’s the setting I use to remove FM carrier hiss and webcarrier hiss.


It depends what the sound to be removed is. Noise Reduction is for steady, similar noise.

If it’s an e-mail ding-dong noise you might be able to ameliorate it considerably with Noise Reduction.

If it’s a song with rock band or an orchestra, Noise Reduction will not make a good job of that.


I know this thread is from 2015, but I had this issue today and I registered just to say that this thread helped solve it.

I was recording my speaker output from my desktop (capturing a radio livestream) when one of my apps played a notification sound that got mixed in. Getting a clean recording of that sound was easy and the invert trick worked crazy well.


Hi! I am new to Audacity and would like any help on removing a notification sound from a recording. I have the notification sound on a separate wav file but cannot figure out how to invert mix it with my main audio file.


I want to remove disturbing whistles, “aaa” s and specific words like “alright” , “actually” etc from the lecture recording. Is it possible?

There’s no automatic way to do that.

You can of course work through the recording, select an “alright”, delete it, then move on to the next one. It’ll take a while to do. Editing a speech recording is usually the part that takes longest when making high quality speech recordings - typically around 10 hours editing for a 1 hour recording.