Just burning time with Python and Graphviz…
There’s actually of course 256 fragment voices. Note that currently the Aux 1 channel receives from individual voices, not from the sum of the Fragment Scheduler. The plan is
to eventually also allow sending from the Fragment channel itself, and of course have more Aux channels.
The Channels process approximately like this (note how I heroically went into the trouble to compensate for DSP induced delays to make the Dry/Wet Mix work nicer 😛 ) :
Fragment voices have a flow approximately as in the following diagram. Note how the fragment overall volume is handled by the scheduler, not the voice
itself, to allow sending into the aux channel pre-fragment volume.