Musicbrainz Picard + PlexAmp is a pretty good solution.

Picard sets _all_ the metadata on the music and PlexAmp uses it to create playlists with the OpenAI API.

Nice, thanks for the rec. I'd seen beets as well which also looks good.

Spotify is definitely the lowest hanging fruit for culling on my subscription list. The API was very much part of the value proposition.

I don't believe it's even that good a deal for artists. I heard the mighty Snoop Dogg makes like USD 40K a year off it or something stupid like that.

Snoop Dogg made USD 40k for one song, which also has multiple songwriters on it. There’s a good breakdown in [1].

[1] https://www.reddit.com/r/interestingasfuck/comments/1ba0ggi/...

Key phrase from that research:

> I'm not sure how much royalties you can expect when you're one of 17 credited songwriters on one song you don't even own which samples a song that also samples songs...I think $45k is pretty damned good.

> I don't believe it's even that good a deal for artists. I heard the mighty Snoop Dogg makes like USD 40K a year off it or something stupid like that.

What would be the non-stupid amount for Snoop Dogg to make off it?

I also want to add that MetaBrainz (the non-profit foundation that MusicBrainz and Picard are the base of) also have ListenBrainz, a music discovery platform that we built because of similar discontent over the way recommendations were going. Also has the advantage of being free and open-source ! (I work with MetaBrainz)

Why?

Iiuc this is just about APIs for the recommendation engine

It’s never been easier to generate recommendations (eg via LLMs and other routes)

The core functionality otherwise remains unchanged in the API

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.

Cool Good to know, I won’t mention it to you again

Maybe you have ondansetron around

The cool thing that's gone now was the "Audio Features" endpoint ( https://developer.spotify.com/documentation/web-api/referenc... ). You could easily get some important values about every song, now you would probably have to run your own analysis for every song you're interested in. That's a lot harder and slower if you don't want to preprocess every single song available.

yup. The ability to make anything interesting with the Spotify API just got flushed.

So happy I never got started on that little dream project that’s been knocking around the back of my head for a couple years…

>The ability to make anything interesting with the Spotify API just got flushed.

I respectfully disagree but ok

Care to explain your disagreement?

Otherwise cool, you prolly wrong but kay

"I respectfully disagree but ok" and nothing else is probably one of the most disrespectful ways to disagree with someone.

Saying an extremely rich API is “useless” on account of several interesting but free APIs (which were probably being abused) being deprecated is absolutely nuts. My intention was to point this out.

I’m not wasting my time at this point explaining to someone what I think is cool about the Spotify API, especially after being flagged. The mob here has already demonstrated immense one-sided thinking.

Being scapegoated for the very thing the mob has done itself is an angering experience.

I mean I don't expect you'd have much of a quality explanation anyways when you're already resorting putting false words in their mouth...

The person you replied to didn't call it useless, they said "The ability to make anything interesting with the Spotify API just got flushed."

Seeing as "several interesting but free APIs" were deprecated (your own words) it's a very reasonable take: most applications of the Spotify application that did anything more than present a basic music player interfaces relied on one or more the deprecated features.

Man you dumber than I thought. You got flagged because you made a low quality comment on a website where quality is expected. This isn’t Reddit where you can post literal garbage and receive zero pushback. and this post is filled to the brim with victimized word salad, calling those who disagree a “one-sided thinking mob” and that you were “scapegoated”. Bro we are not mad at you in place of Spotify, we said your comment was low quality and not worth showing.

Truth is you don’t have an explanation. Because there is no justification for closing down API endpoints like this. You’re just bootlicking Spotify despite their unreasonable actions, and got mad when others didn’t prostrate themselves in front of your corporate overlord of choice.

Are LLMs actually good at music recommendation?

No, they're absolutely garbage at it. I don't even understand the thought to use LLMs in the first place. And even if they weren't garbage the whole point of a music recommendation algorithm is surface music that wouldn't be in the training set so you need a way to recall likely matches at which point you've built a recommen engine.

> I don't even understand the thought to use LLMs in the first place.

You know how people believe whatever they read, hear, and watch even though it might not be true? Well an LLM is something people read and to get over the hurdle of whether something might be true or good, you simply embrace it and ignore that it could ever be wrong. I don’t get it either as I get upset when I find out a source is mostly wrong.

> Are LLMs actually good at music recommendation?

As far as I can tell, the only thing actually any good at music recommendation is (some) humans.

The Apple Music recommendations have been pretty good for me, particularly when autoplaying based on a song/playlist. Which came as a surprise because I always found the Spotify ones to be pretty useless.

Especially with the autoplay functionality on Spotify, after a while it would always throw in some songs that just did not fit in at all with the rest, and it made for a jarring experience. I've never really had that happen with Apple Music.

I’ve tried apples , agree it does less egregious stuff than Spotify but doesn’t really meet the bar for “good” in my experience. This stuff is probably really dependent on what you are looking for I imagine .

Spotify's algorithm has worked very well for me in the past.

I'm pretty sure it's not LLM based though, but rather domain specific, or possibly just a simple recommendation engine ("people who like x also often enjoy y").

Not my experience at all, fwiw.

Yeah but didn’t Spotify kill any community features long ago?

YouTubes algorithm (especially before they nerfed it) is often outstanding at it, pulling undiscovered gems that completely match your tastes.

So you're here telling people who were actually using these APIs that we're wrong to be upset, because LLMs? Awesome, super helpful, thanks

LLMs require data, as I'm sure you know. This is locking up what was previously an interesting source of data, which undermines your argument over the long term

Slippery slope fallacy

Can you do anything aside from low-effort shitposting?

Disagreeing with folks on this thread is not shitposting. Please don’t be an asshole and accuse me of “shitposting” merely on account our disagreement.

Great, LLM recommendations. A list that randomly spits out the names of songs that don't exist.

Just what I want in a recommendation engine - to sift through garbage nonsense.

> only benefit Spotify provides over my MP3 collection is that it's better organised.

This seems like pretty low bar.