There was a discussion yesterday on SG about the RIAA’s new campaign to require the labeling of music in which the recordings are provided by AI, but not when the lyrics or composition are provided by AI. This is why I am against the labeling, because I believe that as with the registration of guns, it is a first step toward attempting to ban, or in this case, deplatform it. More about that at AI Central, as well as a new mix of ONCE THERE WAS SORROW in which I discovered that country rock with a double-lick drum works surprisingly well.
And in passing, I think it’s worth addressing the posturing of those who insist they are very special members of the 3 percent who can reliably distinguish between AI music and organic music, which I suspect has a lot more to do with the fact that since they avoid listening to it, they have no idea how much it has changed over the last 18 months, or realize that one cannot judge the technology by a single application of it; as much as Suno’s voices have improved, it has never been the AI system with the best vocals as its popularity stems from its compositional capabilities, not the vocals.
Soundwaves are soundwaves. They’re not magic infused with bits of the human soul. The “too perfect” problem has been solved before, multiple times, by audio engineers. It will be solved again. It is safe to assume that within 36 months, and probably within 18, even audio engineers will no longer be able to distinguish between human and AI vocals.