I used to use last.fm quite a bit. I haven't checked in a long time, but there was a great "similar artists" section on each artist page. It made finding new tunes really easy.
Other than that, I try to keep up with what my friends are listening to.
I'm really bummed that the streaming services never integrated with Last.FM. I have years of scrobbling on my account from ~2006 thru 2013 and the recommendations were always amazing.
Recommendation algorithms are mostly garbage, I find most of it through RYM by browsing the charts / clicking on people's profiles and seeing their highly rated albums. Human curated > machine curated still.
I wholeheartedly agree with human curation especially on the long term outlook. Sure the algorithm gives you a bunch of new stuff in the first few months but after a while you’re just stuck in a minima. Also I think my taste evolves quickly enough but algorithms box you in to what your preferences were.
I may not like everything in a human curated list but I’m exposed to more things that can have sudden jumps from my current localization.
Sadly, I use YouTube Music, payment for which also includes YouTube Premium. I've been wanting to find an alternative, but too many creators I enjoy following are only present on YouTube.
Spotifys algorithm doesn't do anything. YouTube is a tiny bit better. But it's very difficult. You need to click around to get out of the algorithm that shows the same thing for six months.
In my experience I can tell that music is discovering me. I don't look for music actively. Sometimes I stumble upon new music I like on YouTube or I hear a song on the radio in a cafe.
Exploring the neighborhood around what I like usually works. This means: see what's released on the same music labels. See which journalists and bloggers wrote about the thing i liked. See what artists the original artist considers to be their peers, or inspiration. And reiterate once you find something nice. For random/distant discovery there are radio shows, can seach tracklists on (e.g.) NTS to see which shows play things up my alley. I really don't like the soundcloud UX but for me their recommendation algorithm is quite good, often recommending very obscure but good, and more importantly, relevant things.
>I make multiple Spotify playlists, each dedicated to a particular type of music
This is the correct way to make use of the so-called 'algorithm'.
I say 'so-called' because I know the algorithm to be dumb/blind. Pick something obscure on a playlist, you will get tour own recommendations back. iTunes is the same. You are gaming the recommended list correctly.
Other peoples lists are even better (when they are good ie thoughtful, lists).
There's a website called Every Noise At Once [0] which contains a list of almost every genre of music in the world. Each genre has a linked Spotify playlist with a few hundred songs of the given genre. And once I'm in that genre, I mostly just let Spotify algorithms do their thing to get me new stuff.
You could say "Spotify's algorithm", but you need to help it. It is more their links between artists that I use, and the general availability of a lot of artists. (Not as many as on YouTube.)
If I have found a lot of great, new music this week, next week's "Discover Weekly" and the albums recommended in "Home" are better.
Most of the music I find is by looking at the tracks I'm already playing: Maybe I only heard one song from an artist, maybe I only ever listened to their old stuff, or maybe I only ever listened to their new stuff. Maybe it's a remix, and the featured artist is the one I like, so I listen to their top songs.
I've learned that whatever one-hit wonder people have -- the song that everyone plays 10 times more often than their second-most popular song -- is often a good choice if you're assessing a musician. But if you like them, it's also likely that your favorite song isn't everyone else's.
I mostly listen to electronic music without vocals, so I can play a lot of random music while I work, and occasionally something sticks out in a good or a bad way.
Last year I spent 65000 minutes listening to music on Spotify.
I make extensive use of the Spotify algorithm to discover the music I want to listen to based on my own idea of what it should sound like. For example if I want some music with a house beat but folky feel to it, then I'll create a playlist and 'seed' it with some house music and some folk music; about 10 tracks total. Then I'll scroll to the very bottom of the playlist and go through the Recommended Songs. Spotify is often pretty good at finding an 'average' of the different songs added to the playlist. I add the songs that fit my expectations to the playlist which further refines the recommendations. After I've added about 10 songs I'll remove the songs I used to seed the recommendations.
Sometimes this method doesn't work because maybe the 'average' you're looking for doesn't exist or Spotify hasn't connected the dots. Sometimes a track in the 'seed' can dominate the recommendations so try removing tracks that seem to skew the results or adding more from the opposite genre.
I also make use of “song radio”, i.e. auto-generate a playlist based on a single song.
I might call your method “playlist radio”, and I don’t use it as often.
However, judging by how many songs on these recommender radios that I’ve already liked, I think that recommendations are somehow always based on a vector.
I never really liked exploring music on Spotify the back catalog was non existent for DJs putting out amazing stuff in the 90s and 00s. You get like 2010 all the way to current year. Worthless.
Do you need algorithmically curated music if you're only interested in music that is decades old, especially DJ remixes? Looking through historic DJ set list or even just listening to new sets from old DJs seems like it would be much faster.
Spotify is great for recommendations of licensed content that is new. I rarely feel there is some small or big hit I'm missing in the genres I like, but it is heavily supplanted with the use of SoundCloud for sets and music that will never legally be licensed and on the platform.
I've been involved in punk/hardcore community for decades now and I've got my sources : friends, blogs, forums, last fm follows... For anything else I listen to the radio, especially fip.fr who expose me to Music I wouldn't think I could love. Also, i mainly listen to albums, never playlists, automated or curated ones. I don't believe in recommendation algos and don't feel the need to get into that way of discovering music and leave that passion to a company. I always loved to dig into stores crates or read reviews and I fear the infinite of choice from streaming platforms or being stuck in a bubble. There's like something "unpure" and Betrayal to how I used to love music in Spotify recommandations and the way it pushes always something new. It may sounds puerile or posh but that's just how I feel, i'm very oldschool and conservative regarding this topic.
I Second FIP, here is the official radio and secondary web radios around one theme (Groove, Jazz, Metal, World Music, New EPs that were published recently etc.) all in one page in free access https://www.radiofrance.fr/fip/titres-diffuses
"Late Risers Club" on WMBR (MIT radio) is pretty good. They're daily 10-12 EST, but archive the shows for 2 weeks. Its less pure "punk" than it was but (Some DJs have more rock like tendencies...), but its still great.
French mainstream radio is good, and FIP is very good on that list. I like to listen to it from time to time to just realize how much eclectic genres there are (I am 'ot really into music, just listening casually).
Another gem is the music played when the national radios are on strike. This is a special curated playlist just for that occasion (or a technical disaster) and the music is really good.
I once switched on France Culture and there was sick music, thought "great they are on strike" (yes, quite an unusual reaction) and the host was back and said "to those who just joined, we are not on strike, just listening to some nice music"
Today some tradies were putting up scaffolding next door to me while I was working at home. They were playing this amazing reggae tune, followed by a playlist of similarly great stuff I’d never heard before. At lunch I went for a walk, and on my way out I went over to ask who put together the playlist. Turns out it was Youtube Music based off the first reggae track. So I figure if you can find that first entrypoint song, that’s different enough from what you usually listen to, the algorithms can take you on quite a journey.
https://www.ohmyrockness.com/ is fantastic, I think because it's personally curated by boots on the ground music fans who make a living researching up and coming New York bands. It has a focus on the NYC scene, but features bios and reviews from bands from all over the world.
I like it because I know the recommendations never come from algorithms, but come organically from other music fans. Rock's the overarching thematic focus, but you'll find hip-hop, jazz, and nearly any genre you could think of on there :)
A good strategy for discovering new music with online radio stations that archive their sets is to search for a song/artist that feels very unique to you (not popular or universally acclaimed). Usually the rest of the set will have music you like, and because it’s made by a human it won’t sound same-y.
I find most new music through NPR’s Tiny Desk series. I discovered IDLES a few months ago, and they’re really all I’ve listen to since. Music is a huge part of my life whether it be writing or playing or listening and I’m struggling to think of another band that has had such a powerful effect on me as IDLES.
125 comments
[ 2.4 ms ] story [ 232 ms ] threadOther than that, I try to keep up with what my friends are listening to.
I may not like everything in a human curated list but I’m exposed to more things that can have sudden jumps from my current localization.
What do you mean by this? I discover a lot of music through Spotify.
My mix for discovery usually goes through:
1. Labels that release artists I already like
2. Recommendations from friends
3. Spotify recommendations algos (Discover Weekly, mixes, radio feature, etc.)
4. Music zines/magazines/blogs/critics.
5. Bandcamp recommendations (not so good but worth to check here and there).
Spotify is pretty good to recommend very similar tracks or artists, never had an issue with it so would be keen to hear what's your issue.
2) contextual discovery - e.g. I like some band and than I discover the previous band of their drummer
3) recommendations from friends, internet radio DJs etc
4) accidentally visiting band's concert
Spotify Discover seems to work pretty well for me.
Yesterday it surfaced an outrun track featuring both guitar and sax solos, which... is pretty on-brand for ethbr0.
Having diverse and varied playlists seems to help? I've had friends complain it doesn't work for them, so I must do something different.
(Doom Flamingo - Domingo's Drive, if anyone's curious https://m.youtube.com/watch?v=uH0IUJtUBek )
This is the correct way to make use of the so-called 'algorithm'.
I say 'so-called' because I know the algorithm to be dumb/blind. Pick something obscure on a playlist, you will get tour own recommendations back. iTunes is the same. You are gaming the recommended list correctly.
Other peoples lists are even better (when they are good ie thoughtful, lists).
[0] https://everynoise.com/
I still find Pandora to be the best for finding new artists, I just create new stations and see what they recommend.
You could say "Spotify's algorithm", but you need to help it. It is more their links between artists that I use, and the general availability of a lot of artists. (Not as many as on YouTube.)
If I have found a lot of great, new music this week, next week's "Discover Weekly" and the albums recommended in "Home" are better.
Most of the music I find is by looking at the tracks I'm already playing: Maybe I only heard one song from an artist, maybe I only ever listened to their old stuff, or maybe I only ever listened to their new stuff. Maybe it's a remix, and the featured artist is the one I like, so I listen to their top songs.
I've learned that whatever one-hit wonder people have -- the song that everyone plays 10 times more often than their second-most popular song -- is often a good choice if you're assessing a musician. But if you like them, it's also likely that your favorite song isn't everyone else's.
I mostly listen to electronic music without vocals, so I can play a lot of random music while I work, and occasionally something sticks out in a good or a bad way.
Last year I spent 65000 minutes listening to music on Spotify.
Sometimes this method doesn't work because maybe the 'average' you're looking for doesn't exist or Spotify hasn't connected the dots. Sometimes a track in the 'seed' can dominate the recommendations so try removing tracks that seem to skew the results or adding more from the opposite genre.
I might call your method “playlist radio”, and I don’t use it as often.
However, judging by how many songs on these recommender radios that I’ve already liked, I think that recommendations are somehow always based on a vector.
Spotify is great for recommendations of licensed content that is new. I rarely feel there is some small or big hit I'm missing in the genres I like, but it is heavily supplanted with the use of SoundCloud for sets and music that will never legally be licensed and on the platform.
https://wmbr.org/cgi-bin/show?id=6785
they post there set lists to "Track-blaster". They have a "listen" button if its within the 2 week archive period.
here is last mondays show: https://track-blaster.com/wmbr/playlist.php?id=47208
Another gem is the music played when the national radios are on strike. This is a special curated playlist just for that occasion (or a technical disaster) and the music is really good.
I once switched on France Culture and there was sick music, thought "great they are on strike" (yes, quite an unusual reaction) and the host was back and said "to those who just joined, we are not on strike, just listening to some nice music"
What's cool is all the 'new' music on them, which are simply B-sides from my favorite groups or unheard tracks from the albums.
I like it because I know the recommendations never come from algorithms, but come organically from other music fans. Rock's the overarching thematic focus, but you'll find hip-hop, jazz, and nearly any genre you could think of on there :)
Now: Listening to what Apple Music puts in front of me. I got lazy.
Mostly https://doyou.world/ and https://www.nts.live/
https://www.hypem.com
If you haven’t already, I’d recommend giving their very first “Welcome” EP a listen.