On the API front, the endpoint that's being killed that was most interesting to me is actually their music analysis one. That was super-nerdy fun to fuck around with, I had a half-finished project on that with an old job. Totally interested in hearing of feature-parity alternatives I can run locally. I'd also thought I'd sometime get around to doing some network analysis with related artists too.

I honestly don't find Spotify's recommendations all that great. I definitely experienced a broadening (perhaps deepening) of my listening early on, but my experience has been that the recommendations are pretty shallow.

I find after throwing together a playlist with some stuff I like, it'll add a few more artists to my mental roster, then nothing. I'll get thrown around in the same loop with the same tunes and artists -- usually from the more famous albums.

I don't want to sound too much like the grouchy aging hipster that I am, but recommendations engines are just one of many ways of discovering music, and I feel like y'know, the old ways were better than just paying some company to do it for me. I'm talking here about being a regular on a local music scene, smoking weed with musicians, trading MP3s on the sneakernet.

Another thing where we just pay some money for "convenience", but are left with some hollow and empty algorithmic imitation of something we once loved.

Your LLM suggestion made me do a little sick in my mouth.