Hello Audacity Experts,

I have only been a user of this excellent piece of software for a few weeks, and as such I don’t know all that much about how it works - so please excuse my potentially stupid ignorance when requesting assistance with the following point…

Can a sample from a song which contains several layers (I.E: melody, bass and drums) in some way be altered so that all of the unwanted elements from this portion of audio are filtered out entirely, leaving only the component that’s actually desired? If not, is it more realistic to expect that only samples taken from a section of the song where there is no significant sound competing with the noise that you want are viable for usage?

Although I’d definitely prefer to continue using Audacity, if this goal can’t be achieved by doing so, would a commercial D.A.W product such as Steinberg’s Cubase have the capabilities to manipulate samples in this advanced manner?

Thank You For Your Time - SIMON

Short answer - It can’t be done. “You can’t un-bake a cake or un-fry and egg, and you can’t un-mix a song”

You can isolate the left & right channels (or the 6 channels in a 5.1 surround recording), and you can completely* remove the phantom “center channel” by subtracting right from left (see the Vocal Remover effect)

And to some extent you can remove the left & right while preserving the center (with special software).

And, to some extent you can filter-out the bass (or filter-out everything except the bass), but in general there is a LOT of overlap with most instruments & vocals covering the same frequency range, and often playing the same notes.

Most modern recordings are multi-track recorded. Each instrument & vocal is recorded on a seperate track. That way, each instrument’s volume can be adjusted & edited seperately, or re-recorded seperately, and different instruments are often recorded at different times. Then it’s “mixed down” to stereo.

If it was recorded in stereo (or mono) this wouldn’t be possible. If it were possible to isolate the sounds, we wouldn’t need multi-track recording.

With a [u]Mixer[/u] (used for recording or live with a PA system). Each input is controlled/adjusted before the signals are mixed.


  • This process may not completely remove the vocals and other centered sounds because some sounds may be slightly off-center or may contain out-of-phase reverb that doesn’t subtract-out.

We can’t take a show apart into individual instruments or voices. Very few software packages can do that and the ones that do don’t always leave a theatrically desirable show behind.

Effect > Vocal Removal can remove everything in the left-right center of a stereo show. That’s a relatively simple arithmetic calculation. It’s not “recognizing” the voices. It also rarely does a very good job.


Okay, thanks a lot to both of you - I assumed that would most likely be the case, but I do genuinely appreciate the clarity it has given me by you two taking the time to help me out all the same.