If the visual was able to run correctly on the empeg (as a "standalone app") and take the audio feed as input, how difficult (aside from time issues) could it be for Cambridge to add it?

It would require that it be rewritten to work inside the framework of the empeg's visual engine.

The visuals on the empeg are not a standalone apps, they're based on sets of internal functions. Someone who writes their own external program would be reinventing a bunch of wheels. So in essence, translating an external app to run within the empeg's visuals framework would be a porting project.