I’ve actually found that I’m able to process two speech streams and one music stream independently from those. At least for me, music seems to have a dedicated sub-processor of sorts.