Comment by exceptione

Noob question: does this only separate instruments, or can it also be used to separate and extract speech from different humans in a conversation?