The myth of 'sound machine', ie the black box that produces music player has ancient origins, but systematically uses over the centuries, as a miraculous object capable of generating sounds ecstatic. Over the past thirty years, the science-fiction accomplices and the high scale integration of electronic devices, this object has changed its status from a legend prototype. The code written to implement algorithms generative music has released the aims of the creation of autonomous musical entities, perceived almost as a 'live'. A primitive stage, perhaps illusory artificial intelligence, which in rand ()% is coupled to the distribution at the request of an online radio. The project, developed by Joe Gilmore and Tom Betts in fact consist of the streaming works generative electronic musicians that make perpetually active transmission even though it is never equal to itself. It is the next stage of use of this type of composition, where the fruits of singular originality of the respective machines software are distributed in sequence. In this way the machine is heard in his compositions through a parametric broadcasting continuous and programmed in cables and network protocols. And so, through the centuries, the 'sound houses' in which each sound is born, imagined by Francis Bacon in his New Atlantis, as well as compositions for nuts codified in precise instructions by Mozart in 'Musikalisches Wurfelspiel' are a public form in this project that exploits the potential amplifiers inherent in TCP / IP packets, to offer indefinitely its cornucopia of sounds.