I mix in discreet quad, with each speaker receiving unique signals (as distinct from using 2 pairs of stereo speakers). I have explored a few ways to do this in Live.
My first method is with sends:
4 sends to 4 speakers: these return channels are routed to 4 mono outputs of an RME Multiface I/O, then out to a mixer, & from there to 4 speakers. These are my main output busses.
The 4 sends are mapped to 4 knobs on a MIDI controller. This way each channel can be manually assigned &/or panned to any & all speakers using the knobs on a couple of Drehbank MIDI controllers.
I have an additional 4 sends for 4 effects returns, each containing a variety of effects. These return channels are then sent via the first 4 sends to the 4 speakers. These returns also have Live’s External Audio Effect to route audio via the RME out to & back from external effects.
Because Live sums stereo source material when sent out through the mono outputs, I also have 4 sends who’s return channels are assigned to stereo outputs: 1/2 panned left, 1/2 panned right; 3/4 panned left, 3/4 panned right. These are used for when I want to maintain the stereo separation of source material.
So each channel has:
- 4 sends/knobs to speakers (A, C, E, G)
- 4 sends/knobs to internal & external effects (B, D, F, H)
- 4 sends/knobs to stereo outputs, can be used dry or with effects (I, J, K, L)
This makes for a very flexible system, given the limitations of Live, & my old PowerBook 1.5GHz laptop (to be replaced in the near future –probably revise some things when I have more power at my disposal & can use more effects).
Since this first system is primarily designed for performance & improv work, it was important that the design was very simple & visual: I can look at the screen & see what speaker(s) each channel is sending sound to. And I can see the exact same info on my MIDI controllers. I can mix without looking at the screen if I choose to because the visual information is identical: 2 columns of 4 knobs per channel: column A,C,E, G for dry; column B, D, F, H for effects. The stereo I, J, K, L channels are currently unmapped to any MIDI controller.
For a number of reasons, I chose to limit my track count to 12. It’s enough to let me build up a fairly thick sonic brew, but simple enough to manage, & fits my various MIDI controllers well.
My second method: sends, racks, chains
Iis an extension of the first method, which includes racks & chains in association with the External Audio Effect. Without going into detail, I can pan audio from speaker to speaker with one knob. But unless I make a ton of chains this is limited panning.
One of the major drawbacks of using this second method is that it is less visual: while I can see the speaker assignments in the MIDI controllers, I can not see them all at once in the Session view, because some speaker assignments are now buried in racks.
I’ve just started exploring this second method, so maybe I can conceive of something more refined after experimenting some more.
Don’t know if any of that pertains to your working methods

, but let me know if you’d like any more info.