Have been making music using ML/AI tools for about as long as I've been on HN. Had a recent comment section on HN here [1] and after that thought it might be worthwhile to write out about the journey in greater detail... from crude LSTM models, to hacking around with Jukebox, and now Suno.
[1]: https://news.ycombinator.com/item?id=45388822
+1 to Suno/genaudio being both fun, and evolving editing tools, but the mental justifications in the blog post are painful. I think you may find musicians bristle at your claim that all art is circular, because they are acutely aware of how sampling, copyright and ip ownership rights work, and legal risks of using recorded material as inputs doesn't fly for commercial music. If these were songs you wrote, youd own the copyright on them. The US legal system says no,you are not making music, its not copywritable due to some of the issues you highlighted in your post.
Further reading: https://help.suno.com/en/articles/2746945