Here's how I figured each one out, I got 8/10 (missed 6 & 7) 1: There was no main line in A, I felt like there was a synth trying to make it's way to the top of the mix but it was some combination of too quiet, to low in pitch, or not a "singable" line 2: The vocal sample for B was very low energy, breathy, and no words were distinguishable, it wouldn't make sense for a human to put this vocal line in a song if they were going to actually put effort into the rest of the song unless the vocal line was a joke 3: A had some random rimshot sounding beats in the middle that would have no logical place for a human to put in 4: Creator explains pretty well, also the vocals sound really choppy and breathy without an easily "singable" phrase that fits well with the rest of the song, sorta the same reasoning as #2 5: A at the start, the lead in to the main phrase is very odd, something that would be appreciated if you were listening very hard but the rest of the song is repetitive so I feel as if it isn't to connect with the people who listen to music that close anyway, this mix of intentions is what told me 6: The instruments used are very weird, there's a flute that sounds very overblown on the top of the mix and then an acoustic guitar line below it just doesn't make sense I suppose, there's also some weird crackles in the recording that don't sound natural 7: It's very subtle but the EQing of the vocals changes a bit randomly, like EQ can change over time just this seemed random and non-cohesive 8: At the beginning there's some random clicks, sorta sounds like a woodblock, they're on beat but they don't add too much to the song and I don't think people would think of adding something like that 9: This one was deduction, I was trying to do all of them picking which one was the AI instead of picking which one was human so we could pick up on more audio cues of AI generated than pick up on what sounds human but the way I figured this one out was that A had very complex vocal samples and too much dissonance, even if you asked AI to really hard you wouldn't be able to get this much dissonance, it's like asking chatGPT to write slurs, it just won't do it even if it wanted to. 10: This one was hard, it doesn't recognize or if it does it doesn't respect the main melody line of the staccato sounding synth in the tenor-ish range, the high soprano atmospheric synths in the higher register make no attempt to be a main line yet they're way louder than the staccato synth.
Wow I got most of them wrong. Melodicaly it does a good job. At 5:09 I just thought if even this vocal is AI generated then I will drop all my papers. Maybe 2 papers down the line it will be able to do that?
Ai does not have feelings so the world will be flooded with more and more garbage music from now on...as if there was not enough garbage made by emotionless people.