Having seen a lot of DIY track separations getting posted in the Stems/Multis section I decided to get a copy of the software and try it out. The algorithms in Music Rebalance extract audio from a track and split them into Vocal, Percussion, Bass and Other output stems.
From what I've experienced the app is good at extracting DIY Acapellas and has a good go at isolating drum tracks in Percussion, but it all goes pear shaped when it tries to extract what it considers to be the Bass track, which invariably ends up as a muddy, almost sub bass track. The residual, Other instruments track is far better.
I tested the app out by feeding in just a bass guitar track from a studio stem set and the algorithms split the source into a muddy bass frequency track and the majority of the information ended up in the Other track.
Whilst RX 8 seems very handy for producing usable acapellas, where artifacts can get lost in the mix it seems pretty hopeless for producing usable stem sets to give separations of backing tracks.
I wonder if other people are having better results with it and might have some tips to share on how to get a Bass track from the algorithm that is listenable on its own?
From what I've experienced the app is good at extracting DIY Acapellas and has a good go at isolating drum tracks in Percussion, but it all goes pear shaped when it tries to extract what it considers to be the Bass track, which invariably ends up as a muddy, almost sub bass track. The residual, Other instruments track is far better.
I tested the app out by feeding in just a bass guitar track from a studio stem set and the algorithms split the source into a muddy bass frequency track and the majority of the information ended up in the Other track.
Whilst RX 8 seems very handy for producing usable acapellas, where artifacts can get lost in the mix it seems pretty hopeless for producing usable stem sets to give separations of backing tracks.
I wonder if other people are having better results with it and might have some tips to share on how to get a Bass track from the algorithm that is listenable on its own?
Last edited: