need help removing music/lyrics from an audio file so that you can hear the background talking in the room.
i can send you the file by email.
That’s generally impossible and your ear is often the best “filter”… If you can’t understand what’s being said software/processing is unlikely to help.
Music covers all (or most) of the audio frequency range and voice requires a big-chunk of the middle of the range… There’s a lot of overlap so you can’t separate one from the other.
If you have a stereo recording you can edit/mute the left & right separately and there are “tricks” for eliminating/isolating the “center”.
“You can’t un-fry an egg or un-bake a cake, and you can’t un-mix audio”.
It’s not impossible, but can be incredibly difficult. However, multiple voices are all mixed very fine if you look at them in the spectrogram and the overlap is just way too much for it to be even remotely close to possible (the only exception is if there are high-pitched backing vocals being played against low backing ones; then it’s doable)…
@DVDdoug You’re wrong when it comes to un-frying eggs 
https://www.cnbc.com/2015/01/23/proteins-and-research-scientists-figure-out-to-unboil-an-egg.html
Whenever you discuss your top secret plans, be sure to turn a radio on in the same room with you. It’s killer difficult to pull your voice out of that.
Oddly, Hollywood didn’t make that up.
Audacity doesn’t do surveillance, law enforcement or conflict resolution. There are companies you can hire to help you.
Koz
@DVDdoug You’re wrong when it comes to un-frying eggs
Someone on Google claims to be able to get the chocolate out of a cake.
Koz
I didn’t really appreciate that snarky response… The fact is this: More and more programs are coming out which try and separate songs out / into approximations of the original tracks, and the tech will only get better with time. If you had to ask me, there will be a time where you will be able to upload a song to a site and process it and have it send very close (and more than usable) “stems” (separated tracks).
that, and the site I linked isn’t some fly-by-night news source… they’re reputable.
Oh, and if you want to be 100% technical; you can unmix a song entirely and have the results be 1:1 copies of the original tracks, but it’s pointless since you’d need the original multi-track sessions to start with. phase inversion is awesome…
Audacity still doesn’t do surveillance, law enforcement or conflict resolution.
Koz
Behold. Response Partitioning.
Limited time to post before having to perform Real Life.
I have no doubt you can do miracles with current technology, particularly if you have access to the original, or close to the original music.
However.
What happens to the forum is people posting trying to pull The Neighbor Dog Barking out of an impossible mash of other sounds, rescuing a conversation/performance out of the “wrong” side of a Skype transmission and irrefutable evidence of partner infidelity. Or it would be irrefutable if we could hear it.
It was and may still be forum policy that other companies are better able to deal with those. This is even past not being able to do the task at all. Audacity can’t split a mixed performance into individual voices, instruments and sounds.
Removing the chocolate is a running joke along with unbaking a cake and unfrying an egg.
Koz