# machine-learning
- p
polymur
10/24/2021, 9:27 PMi mean - p
polymur
10/24/2021, 9:27 PMif i don't do it, someone else will - p
polymur
10/24/2021, 9:27 PMso - p
polymur
10/24/2021, 9:28 PMi might as well just friggin do it - h
hecko
10/24/2021, 9:28 PMwell for that i don't think you want a recurrent neural net - h
hecko
10/24/2021, 9:28 PMor maybe you do idk - p
polymur
10/24/2021, 9:28 PMno? - h
hecko
10/24/2021, 9:28 PMhold on i remember some prior art by google - h
hecko
10/24/2021, 9:28 PM - p
polymur
10/24/2021, 9:28 PMah yeah i've seen that - p
polymur
10/24/2021, 9:29 PManyway - h
hecko
10/24/2021, 9:29 PMthis does use wavenet - p
polymur
10/24/2021, 9:29 PMgooot it - p
polymur
10/24/2021, 9:29 PMi think the biggest difference might be like...audio is "serial" data where something like an image is "parallel" - h
hecko
10/24/2021, 9:30 PMyeah i get what you mean - p
polymur
10/24/2021, 9:30 PMand i get how a spectral plot would be a parallell analysis - p
polymur
10/24/2021, 9:30 PMbut then again, most producers use uncompressed formats as opposed to compressed ones - h
hecko
10/24/2021, 9:31 PM(it fails) - p
polymur
10/24/2021, 9:31 PMoh shit - p
polymur
10/24/2021, 9:31 PMi meant to watch this - p
polymur
10/24/2021, 9:31 PMimma check it out now - p
polymur
10/24/2021, 9:31 PMalso fwiw midi is a LOT easier to work with than audio - p
polymur
10/24/2021, 9:31 PMbut with midi you're limited to the sounds of the software you already have - h
hecko
10/24/2021, 9:31 PMoh yeah absolutely - h
hecko
10/24/2021, 9:31 PMthis is with spectrograms though - p
polymur
10/24/2021, 9:32 PMooo - p
polymur
10/24/2021, 9:32 PMgod lol - p
polymur
10/24/2021, 9:32 PM"ai-generated" edm reminds me of a scratch program i made when i was like 13 - p
polymur
10/24/2021, 9:32 PM(i eventually quit using scratch bc its ABSOLUTELY USELESS for midi sequencing)