From AJ:

AUDYA doesn't remember sound but location in memory where a sound is loaded. As an example, picture AUDYA's SOUND RAM as such with the various voices loaded in

LOCATION : VOICE
001 : Choir-Ahh
002 : Sax_Down
003 : Pad synth
004 : Achapella Voice (Low)
etc ....

When you edit a style and request it uses voice Choir-Ahh, what gets associated with the style is LOCATION 001 and not the actual sound file itself.

When you edit a PROGAM VOICE and in it, you use a voice (e.g Pad synth), what the Program stores is location 003, which is associated to Pad Synth. Here is the delima .... switch the voices around and your Program voice will now have the new voice loaded on start up (e.g. Sax_Down - if this is now the voice in location 003). This is more disasterous for a style!