Comment by butlike
7 hours ago
I feel like audio-level heuristics will be easier, but ultimately who's to say?
> Generative models synthesize sound mathematically. These synthesis methods leave unnatural dips, specific spectral noise profiles, or phase alignments that rarely occur in real, human-recorded audio
Then the slop merchants will simply move to controlling a DAW with AI and use the same software synths that everyone else does. It's a little more involved and slower, but far from hard.
Ultimately this isn't really solvable without a way of marking audio with a verifiable signature that it was produced by a specific human, with some kind of reputation algorithm.
This is a totally fascist musical bias -- I have been using spectral techniques in music since the late 1980s. It has been common for decades. "Unnatural" spectral packet distortion is a component of a wide-breadth of existing music that pre-dates modern generative AI. I am confident that the false positives will be overwhelming and unfair to many artists. Such a cowardly and lossy solution.