Surely, they can adapt the '3Band' waveform for stems.
Even though Algoriddim and Virtual DJ WON'T ADMIT IT, this AI audio signal separation is based on the OPEN SOURCE 'Spleeter' by Deezer Research. It's so obvious because the results are the same in that you can hear the same ARTIFACTING going on.
With that said, rekordbox 6.0.2 has vocal detection, and they should also be able to adapt the '3Band' waveform for stems. Hoping they're just waiting until they can improve it. Also, hoping they won't charge for it like the vocal detection (those @zzh#les).