July 29ths
7/29/2009
On alternate tunings: When you play in strange tunings over long periods, standard tuning becomes strange, and seems new again when you come back to it. The notion of something being ‘standard’ implies that it was once abandoned and returned to with a new perspective. It reminds me of something Brian Eno once said about going to an extreme then retreating to a more traditional approach. That’s the payoff of experimentation.
[The musical equivalent is the polyrhythm such as 2 against 3 or 3 against 4, or isomelodies and isorhythms].
Black-out lyric and the rhythms in the words:
7/29/2023
7/29/2023
7/29/2024
7/29/2025
If you do both manual creativity and AI-assisted creativity, it’s probably a better idea to make something manually first, and then run it through AI and see what it can come up with. The problem I’ve had, especially with music, is that the AI-generated music tends to create a persistent ear worm that gets in the way of internal hearing. In visual art, you could generate something to use as a photograph to paint from. So instead of going out and photographing something, you could describe something and then work from that. But if you generate a photograph and then paint it, is it really photorealism or is it “AI realism”? Also interesting that once human and machine are in a feedback loop you can take a photo of the painting you made from the AI generation then feed that back into AI, paint it, photograph that, etc, and it would all have the same seed or parentage. The fact that something becomes fungible is mostly a bad thing for creativity, but after 25 years since the advent of the internet, the seeds are in big silos and it is difficult to find the original seeds that would have someone’s name on them. This is the issue with data sets: That guitar solo is probably an amalgam of many stems and it would be difficult to claim ownership on a few notes, unless it could be forensically analyzed, which is easy with AI. So AI can check other AIs.
Comments
Post a Comment