Does anybody know how this works? As somebody with a serious interest in signal processing, machine learning, and audio, I'm genuinely interested. Have Adobe published anything about the technology behind this?