This applies to later versions that you have installed the OpenVINO AI effects for Audacity in.
This one does not bring sounds out of the noise exactly, but I have tried it on a few things and used with some judgment, it can be used on our file and as an extra bonus, if you have aging ears, it can also work wonders for things like tv, movies etc,
The gist of the idea is this; Make a copy of your audio file in audacity. Unless it has a really stellar stereo sound field, mix both the copies down to mono, so now you have two mono tracks that are identical. Unless the tracks also have spectacular fidelity you also may want to resample at least one of them to 16Khz. The ai noise reduction may do this anyway as it is kind of an ai standard.
Take a or the track you re-sampled and tag the whole track and run it through the AI noise reduction effect, When it is done use the channel gain control to turn the gain on that channel most of the way down. Pop your phones on and start both tracks playing and ff to where there is some vocal action, and bring the noise reduced gain up until the voices just pop nicely above the background.
You can play the file like this or if you want to save it, take a look at how far down the gain is on the channel, and tag all the noise reduced channel, and use the amplify effect to reduce the gain down by that amount, and than mix those two channels into a new one and export that.
To me, the AI noise reduction is downright spooky in it's ability to get rid or noises and music, but quiet spots are just too quiet. And the files loos almost all of their original feel when ran through it. However, it is wonderful to be able to give the vocals just a bit of a boost. It keeps the ambiance of the file but makes it presently more listenable. Also if your ears are older and things like the dialog on tv and moves gets lost in the sea of background music and sound effects, this will let you pull the vocals out a fair amount before it starts to sound unnatural, and it sounds nothing like boosting the midrange eq, To me that usually sounds truly awful.