Children raised in a multilingual environment learn all the languages that they are exposed to with no effort. Does the same thing happen if a child has only indirect contact with a language? For example, if the child is only exposed to music, TV and radio programs in another language, is it possible to acquire a native competence in that language?
Another example of this might be the recent acquisition of rather good English by large numbers of people in Scandinavia. In the case in Iceland, initially Danish was taught at schools, but that didn't lead to widespread Danish fluency (at least not to the degree you see English fluency now). After independence they switched to English as the most important second language, with much the same sort of schooling and contact with foreigners (i.e. not much), there is widespread English fluency in Iceland. I suppose some of that could be attributed to children hearing English in mass media.
There was a study (Mothers' speech in three social classes) done in the 1970s where Dutch speaking children in the Netherlands were exposed to German children's television programming for several hours a day with no other German input. These children acquired no German. Likewise, deaf parents used to get their hearing children to watch spoken language programming to give them spoken language skills and these children didn't learn anything. So children can't learn a second language (or, technically, a second first language) without rich interactional input that isn't television.
Adults can learn from television or CDs as long as they're geared towards language learning. That's because adults — and older children — are already literate and have had experience constantly learning things.
Of course, children and adults can learn a second language perfectly well without native speakers — but with near native speakers as teachers. The majority of English teachers in Europe are rarely "native speakers" in a true sense of the word but still teach the language very well.
I can't claim to be an authority on the subject, but believe that the answer varies with the medium.
The way that any language is learned is by establishing a common context that contains an understood meaning and then associating linguistic data with that meaning.
If the new language is only ever heard in isolation without any context, there is no way to learn it. Therefore, if the language is only ever heard on the Radio or in music, there is no way to learn it. If the Radio program is multilingual were a given sentence or speech is translated into a language that the listener already understands, that is different because you now have a context upon which to build meaning.
TV and Videos on the other hand are different. There you have a context to be able to extract a certain amount of meaning from the sounds. A movie could be entirely in a single language but you have the context of the scene etc upon which to build a shared meaning.
Just speaking from anecdotal evidence and personal experience, I have a friend who grew up in a 100% German-speaking environment, but who seems to have acquired English from American children's cartoons. So I would say that yes, it is possible.
That said, how is that contact any less 'direct' than hearing actual people speaking? Language is language, is it not?
The situation in Malta is interesting, supposedly a lot of children learn Italian through mass media. Because of the relative size of the populations, this is different than the Dutch/German case LaurenG mentioned. However, much of the Maltese vocabulary is quite close to Italian already.
A similar situation exists in Scandinavia with English. Almost all television shows, movies and music imported from the anglosphere, and not dubbed as in other countries with larger population. Many Scandinavians find themselves very good at understanding both written and spoken English, but less proficient in writing and especially speaking it.