I've been using this for a little over a year and like it a lot but it lacks the music discovery features of commercial platforms (which becomes important when you have terabytes of music) and takes a lot of effort to organize music and create play lists. The music onboarding process typically starts with organizing metadata with Musicbrainz Picard then import the collection into Funkwhale's DB via a cli admin tool. I have funkwhale hosted at home attached to my own music collection. There is an unofficial mobile client called Otter that makes listening to music on Android very pleasent. Admin overhead aside, this program has become a part of my daily life and I greatly appreciate the developers efforts.
I cannot upvote Funkwhale enough. I've had the need for something like this for a long long time now. A decade ago I was lucky to have my own desktop computer (before it was a shared one with my brother, before that I had just my parent's from time to time). Nowadays, between the smarphone, the work laptop, my desktop and my laptop I cannot possibly have all my music available all the time. And I don't like using spotify where I don't feel I own the songs and they get dropped from my lists. I used to be a fan of grooveshark, but that went down. Now I can have my pod and share it with my friends? On my own server? Accessible everywhere? I'm in.
> There is an unofficial mobile client called Otter that makes listening to music on Android very pleasent
Is there some reason servers like this require a special client? Can't you just provide URLs to m3u8s which in turn have URLs to mp3s? Is it just that there isn't an agreed-upon protocol for listing directories? Or maybe auth concerns?
And maybe for creating playlists? And maybe for modifying playlists? And maybe for searching for songs/albums/artists? And maybe for modifying metadata?
Considering all these features that needs to be supported it makes sense IMO that it's a specific protocol.
Which music service's discovery features actually work? For me the only one that worked was Google Play Music. The rest have been total and utter crap at discovery.
Playlists:
Discover Weekly tends to result in ~3-5 (new to me) musicians/week that I hadn't heard of that I would actually listen to. That's a pretty high ratio IMO.
Release Radar tends to result in ~1 (new to me)/week. Granted, it's supposedly mostly ones I listen to, but still has several I've never heard of.
Daily Mix 1-6 are a mixed bag and sometimes result in something new, but mostly just things I like (and may have forgotten about too).
Discover Weekly was rather boring for me. It was about 6 weeks before I heard a song I hadn't heard before and only one musician I hadn't heard before in the 3 months I used it.
I had much better success with Pandora when I tried it, but it still wasn't exactly deep cuts.
Maybe that's just because I'm too young, but most of my music discovery since I was 16 maybe was done through spotify, so spotify knows pretty much exactly what music I know and what music I don't know.
If by Music Discovery we mean finding music that's new to us as apposed to finding particular music we already have / own, I have two main sources:
* Freeform radio stations (preferably with live playlists): WMFO, WFMU are my favorites. Even people who are immersed in music can't help to hear something new every hour. For me, it's a constant wave of new-to-me music. Many free form radio stations are also layering tracks, interviews, noise, and other audio treats that make for unique experiences that may never (or should never, haha) happen again. Just under free form radio there are countless excellent LPFM and college stations around the country - Hollow Earth Radio, nearly every college radio station from Boston to Milford PA.
* I also use Bandcamp for getting deeper into a genre or trying out new ones. They write up articles that profile maybe a dozen artists that represent the boundaries of a style - whether you read them through or just listen, it's an amazing value. Easily on the level of what the New York Times does for classical music. Bandcamp is obviously growing like a weird and wonderful weed the last year - I would really like them to add a few more features for building random playlists within a few criteria.
Spotify is outstanding in this regard. "Go to radio" from any given track or album; works every time. For even more passive discovery, their "discover weekly" and "release radar" do a good job of surfacing things too.
"discover weekly" was okay, but it is now completely broken to me. It suggests 80-90% of songs I already downloaded, and the remainder are always the same. Also discover weekly stops after, what, 2-3hrs of listening. Then the list is done and apparently Spotify thinks there's no more to discover.
When Pandora first did discovery based on the attributes of the music you're listening to, it was amazing.
Then it became trendy to provide discovery based on what other people who listen similar things like. Not so amazing ever since. Not _bad_... just not great. I haven't found any (even GPM) that do a good job at pulling together suggestions that fit into my eclectic listening habits.
Spotify is quite great at the Daily Mixes (I don't check the other playlists that much).
I go through phases where I deep dive into genre's. Each Daily Mix ends up representing one of the genre's I've been listening to lately pulling in music that I like and other songs that I may like.
That being said, that may be due to the fact that I deep dive into genres that don't have that much of a cross over, e.g. Japanese Hip-Hop & Lo-fi beats, R&B, 90's indie rock, etc. Still, the Daily Mixes are a great way to listen to music I like separated by genre.
Last.fm's discovery feature is pretty neat too. I don't use it as much as Spotify's because I don't listen to music on Last.fm but I think the key feature of Last.fm over other music discovery tools is that it has a profile for many artists.
Spotify, Apple Music, etc. are limited by what music is on their platform. If an artist isn't on Spotify, then they won't have a profile. Last.fm isn't limited that way so you can find even more artists, including artists that may be more underground or niche.
>The rest have been total and utter crap at discovery.
Strongly agree that music discovery is, in some sense, generally 'broken' across most platforms and not very good. I have been deeply dissatisfied with just about every system. I found Google Play music to be okay.
My best luck these days is what I would call a "brute force" search through record labels, last.fm similar artists, bandcamp pages for genres, sputnik listings for particular genres, etc.
It's very hit or miss, but I feel like it leads to me to my occasional lightning strikes, which are what I really want. These discoveries are quite different from the guesses put forward my recommendation engines, which seem to smooth out the interesting edges and signatures of personality and gradually draws toward a lowest common denominator, with no lightning strikes.
I have been reasonably happy with Spotify’s discovery. It hasn’t been impressive as YouTube recommendations, but it’s still the primary way I find new music. Their generated “Discovery Weekly” playlist is a favorite of mine.
I personally enjoy Spotify discovery a lot, the "Discover Weekly" and "Release Radar" playlists made me discover some bands I would have otherwise missed.
GPM worked very well for me and I've used it since the beginning; it was great for new releases and finding new music I like. I ditched it recently due to the YouTube music thing and moved to Spotify. I find im always listening to the same playlist because I just can't find anything I want to listen to on Spotify unless I search explicitly.
I think it depends in personal taste. Spotify and amazon give me mostly boring mainstream ... . The only discovery algorithm i keep enjoying is pandora. due to semi effective geoblocking in my vountry, I, however , hesitate to go for the ad-free subscription.
> Which music service's discovery features actually work?
It depends on what you mean by "work".
Discovery capabilities have certainly gotten vastly better. 10+ years ago, the only decent one was the now (effectively) defunct Last.fm. These days, they're all pretty good. Spotify, pandora, google music, and now youtube music will do a good job of giving you recommendations based strictly on what you've been cue-ing up.
But the recommendations from these services are the equivalent of going into a record store and getting advice from a dim-witted and disinterested employee. You'll get all the obvious stuff, maybe things you forgot about, and if you happen to like popular stuff the recommendations will work OK. But you won't get challenging, provocative recommendations that expand your taste. You'll get cloying recommendations that try to cater to your taste like it was a static attribute. Oh, yeah, and there's "the surveillance capitalism thing" which happens to be the centerpiece of all these services. Is that a problem? Yes.
The best "discovery algorithm" is still HUMAN BEINGS.
Your experience is vastly different than mine. Youtube music seems to be recommending nothing but what's popular. Justin Bieber is being recommending to me. I've never listed to him or anything remotely related.
No good recommendations on any of the others.
Maybe you have a different definition of "good"
Good to me means "sounds similar and in the same genre as what I'm currently listening to". It does not mean "people who liked this song also liked that song"
> It does not mean "people who liked this song also liked that song"
Actually, yes, it does (and many other things too), but to be fair "good" is a highly subjective judgment which is going to be different for everyone.
I don't think, at this point in time, we have recommendation engines that can do much more than fling out recommendations based on an unknown convolution of your listening history combined with music meta-data combined with social network data and a mix of paid stuff courtesy of your surveillance capitalism purveyor.
I know it's possible to capture some characteristics from the music track itself, like bpm (perhaps usable for EDM DJ's?). The "holy grail" would be to have a system that can truly assess the nature of a piece of music based on audio and use it make "interesting" and non-obvious recommendations. We are very far from doing that in software, but humans are still very good at it.
The problem with "people who liked this song also liked that song" is that very often I don't want to listen to that song now even if it's something I really love.
If I'm listening to ambient music I don't suddenly want to be ambushed by something uptempo. If I'm listening to e.g. Debussy, you might be excused for suggesting something vaguely new age in a similar mood and tempo, but certainly not rock.
Another problem is that you can't just take raw overlap in tastes, because some people like "everything", and the fact their tastes overlap with mine does not mean I'll like everything else they like.
I've yet to hear a recommendation system that chooses music I want to listen to reliably enough that I can generally stand to listen to them for more than a few songs at a time without it turning into an endless annoying sequence of skipping.
Respecting genre (segues need to be gradual, if at all), respecting mood and tempo needs to come first. Then you can consider what others who likes the same songs within those constraints also likes within those constraints. Honestly if I have to choose between personalised recommendation and precise control of genre and mood/tempo, I'd take genre and mood/tempo over personalisation any day.
Another pet peeve of mine is lack of visibility into how to teach a system what I want. E.g. if I dislike or skip a song, will it get that it doesn't fit my current mood or what I want to listen to now, or will it wrongly infer I don't like the song at all?
Sometimes it feels as if the people designing these systems don't use them.
> you can't just take raw overlap in tastes, because some people like "everything", and the fact their tastes overlap with mine does not mean I'll like everything else they like.
I think that these recommendation systems, especially youtube's, are much more nuanced than you're suggesting. And I say this as someone who finds these recommendation systems lacking and of limited utility compared to discovery from reading Pitchfork, for example. They're trying to balance quite a lot of inputs and actually somehow make money from it.
Also, "genre" and "mood" of a piece of music are not easy to define, let alone measure. You're asking for a lot and I don't think what you're asking for is realistic from a piece of software.
I'm sure they are more nuanced in that of course they will take into account differences in breadth etc. to some extent, but they are clearly not nuanced enough, because they barely work.
> Also, "genre" and "mood" of a piece of music are not easy to define, let alone measure. You're asking for a lot and I don't think what you're asking for is realistic from a piece of software.
Yet I think if you asked a bunch of humans to attach labels to tracks they like, you'd find they overlap closely enough.
I base that on the fact that while many will struggle to identify niche sub-genres or discriminate between genres they don't listen to, people certainly have a relatively large shared idea about major classifications.
What is more: You'd be able to find the labels of people who usually label the same way as you.
And it doesn't need to be precise. It just needs to better.
DJs should be in your list -- downloading tracklists from sets that I like is my #1 method for discovering new music
Find a couple DJs you like, download everything you can from their soundcloud or mixcloud, then give it a listen. Keeping Shazam close by is also helpful
