The audio won’t bleed between channels unless the algorithm is designed to do that.

For example, my mixing algorithms blend tracks, by design.

In the absense of signal processing to blend channels, there is no crosstalk between the channels.