Any other/additional method to remove voice from stereo audio

Windows 10, Audacity 2.3.3

I am using “Vocal Reduction & isolation” effect to remove voice from anime audio and get the background music.
The audio tracks are demuxed from DVD and PCM.
What bothers me most is that after applying the above effect, some instruments from the center get removed along with the vocals, like whole Bass guitar, Flutes, and parts of all instruments. This produces kind of a hollow sound with almost nothing in center.

  1. I usually have to remove vocals ranging from 120Hz - 16000Hz. Yes, below 16000, remnants of vocals can be heard like high pitch static sounds.
  2. And I use 0.75 Strength, not 1.00.
    Does anyone know of any other method or additional method to retain the center-panned instruments but not the vocals?
    I have read everywhere that it’s not possible, articles always give an example of a bread from which the yeast or flour CANNOT be separated.
    But technology is advancing, so if anyone knows anything, please help me.
    Thank you.

I have read everywhere that it’s not possible,articles always give an example of a bread from which the yeast or flour CANNOT be separated.

That’s generally true. I always say, “You can’t un-fry an egg or un-bake a cake, and you can’t un-mix audio.”

There are a couple of applications that claim they can do it. I found [u]XTRAX STEMS[/u] and [u]DeMIX[/u]. Neither of these is free.


The inability to un-mix is the reason modern recordings are multi-track recorded… So the individual tracks can be manipulated or re-recorded before mixing. If you could un-mix, you could record the whole band at once in mono with a single microphone, then un-mix and re-mix, making a new stereo (or surround) recording etc.

It’s also the reason for multiple microphones on a stage… So all of the individual elements can be separately controlled & mixed with a mixer. And, multiple close-microphones minimize noise pickup and feedback which is also impossible to “un mix” from the audio. Even before multitrack recording, mixers were used with multiple microphones to mix “live” in the studio.

If we could "un-mix"noise that would be great! Noise is pretty-much the “final problem” in home-recording and soundproof studios are the main reason pro recordings are better than home recordings. It would also make movie production a lot easier… Virtually all movie dialog is re-recorded in the studio.

And, it’s the reason that Dolby Digital Surround (with 5 or 7 separate tracks) works so much better than the old Dolby Pro Logic where the surround information was “matrixed” and mixed into 2-channels. (Pro Logic was done with “phase tricks” similar to how vocal removal works with some tricks to “steer” the audio. And, there is a Pro Logic music-matrix mode that does traditional vocal removal in the rear channels without steering which would mess-up regular stereo music.)