Let the user take a photo of themselves, and then prompt them to record a series of soundfs. They can record as many as they want, but prompt them as to the types of things which will work well (e.g. "Ooooh", "Wow", "Interesting", "Tell me more"). These contacts will then be added to the main list of contacts (see #2).

Once a user has added their own contacts, they may no longer wish to see the built in ones. In this case, perhaps we can allow them to hide the default contacts once they've added at least one other.