I’ve been using LastFM for nearly two decades now. First of all, having personal listening statistics is kind of fun. It might be not for everybody, but it’s nice to see which albums are your most played over a year or what you listened to back in 2015, how your favorite artists changed, which album really vibed with you and so on.
Second, you can get really good recommendations for new music when you have a larger user base and are running into a smaller genres. So just like Amazon’s and people who bought this product also bought that product for music. So people who listen to Britney Spiels also like to listen to Christina Aguilera. That might be obvious for you, but it’s totally interesting if you go down some of these genres and if you want to explore them.
And on a broader scale, listening data is quite valuable to create a good music service. So if somebody never heard of a band called Deep Purple and wants to change that, there might be this one song everybody knows from Deep Purple. And this is, of course, the most popular, but how do you find out that this is the most popular? So if you have your own Jellyfin installation, you load in several albums of Deep Purple, but you need some data source to tell you that ‘smoke on the water’ is that famous song from Deep Purple that everybody’s listening to.
ChunkMcHorkle@lemmy.world 3 weeks ago
Same here. I love that shit. My mood is the algorithm. I still occasionally get new stuff, but from other sources I happen to see or hear, like a Netflix show that has it in the background or a musician’s personal recommendation in an interview, and I go look it up manually. But even if I never got anything new, I already have more music than I could easily listen to in a lifetime that I already know I liked at least once.
I’ve tried streaming sources, but it never hits right. This way, where I am specifically picking the artist or album, it’s always right, always fresh, and I’m always listening to something I want to hear.
MangoCats@feddit.it 3 weeks ago
That’s a remarkable level of effort, these days. Yes, I know, it’s trivial compared to pulling vinyl from the sleeve and flipping it every 20 minutes the way I used to before 1985, but… I prefer to put in my music effort with focus, and let a mix algorithm surprise me when I’m not in “music picking mode.” To me, it’s much more enjoyable to hear a song I like that I wasn’t expecting than it is to think about it, navigate the list of thousands to find my pick, and then hear the thing I was thinking of.
HeyThisIsntTheYMCA@lemmy.world 3 weeks ago
I have yet to find an algorithm that has figured out what I like about music. So I curate my own collection. Only way it works for me.
MangoCats@feddit.it 2 weeks ago
What works for me is to have a pool of thousands of “songs I like” - but then you’ve got the mood problem: Metallica or Sarah McLaughlin? That’s what AcousticBrainz was good at: picking through my collection for similar songs and playing that “mood” from the pool of songs I’ve already indicated I like by including them in the available list to choose from.
Where it excelled was at finding the outliers, like the relatively quiet Metallica song that fits with the current set.
ChunkMcHorkle@lemmy.world 3 weeks ago
It’s all good. I think it has a lot more to do with accommodating one’s own brain, and how we individually categorize and enjoy our listening, than with specifics of music like genre/artist/album/track.
For myself, I almost always have some tune or another out of nowhere running through my head, so when I choose something to listen to, I am either picking with or against what’s already playing. So if I tune in to the music that’s already playing, I can see associated choices that are the same, similar, or completely unrelated on a superficial level, but my brain has linked them all somehow. Any of those choices, if I put them on, will satisfy because my brain is already playing one and mentally I’m already there.
I think the reasons algorithms never work for me is because no one could ever follow that, much less predict it. Even I can’t. Instead I’ve learned to simply accommodate it.
MangoCats@feddit.it 2 weeks ago
AcousticBrainz has (had?) a bunch of dimensional measures of various qualities of a song. How I used it was to first define a set of maybe 4 to 8 songs to “set the mood” and then pick a list of a few hundred songs that were “closest to” those songs in all the AB dimensional measures, pre-filtering out artists and songs recently played. Then - the final step was to sort the remaining candidates by their similarity to the songs most recently played. It was still random within that list, but a weighted random with the most similar (by AB measurements) songs most likely to be queued up next.