Choir of voibots (voice bots) interacting in groups with one another. A groups sings a sequence, another group replies.
There are 3 groups, although not easy to identify (they have slightly different vocal tracts). The difference of these voibot from humans is that they sing either breathing in or breathing out. This sequence goes for ages, as they evolve their own music (kind of learn by improvising). I cut 1:45 mins for you here