The band I’m using playback/sample tracks for is like this: we have drums, bass, guitar, vocals live. Then there are certain things that can’t be played by an instrument, sounddesign elements, reversed stuff, special fx etc.
The songs are finished album productions. So I opened the projects, took out anything we play live. The ‘complement’ of the the songs remains and consists of different things described above (different from song to song). Instead of bouncing that all toghether to a simple stereo playback, I sorted it into groups (stems) and exported them individually. Without question this can be time consuming.
The result is a bunch of musically sorted backing tracks that - played at unity gain (0 db on their channels’ fader) - sums up pretty exactly to what it is on the album production minus the live elements. I have a little bit more control now of how loud the different elements need to be for a live situation.
Compared to an album production, where everything is well controllable, live performing has completely differrent dynamics. Here we’re dealing with many more factors - venue size, local PA system, musicians’ mood-of-the-day etc. The multi-stem playback allows to adjust accordingly. I primarily do it via event gains (leaving all channel faders at unity). My live Cubase project sums that to a stereo output, feeding the live mixer - to keep it as simple as possible on that side.
There’s a click track of course, an additional ‘signal’ track, a ‘tuning’ track (some songs begin with vocals, singer needs a note to be spot on). For navigating through the show I use the marker track and the short cuts ‘next’ and ‘previous’.
To get an idea, here’s a video. Just mute anything your brain can identify as drums/bass/guitar/lead vocals to know what’s playing out of Cubase