As a small artist living in quite a tiny niche - didgeridoo music - I have to say that the requirement for having your music on spotify is a no-go. Artists like myself don't benefit from spotify, I don't think, it's a net loss and you join the ranks of people with questionable actions (cf., e.g., spotify-AI-generated content). BC and/or YT would be viable alternatives for those who actually need discovery IMO.
E.g., here's some unexpected percussive organic trance for you that my collaborators and I put on BC: https://dojorich.bandcamp.com/album/tubecolab (I'm Freddie Veggie, btw). Music of that scope doesn't "perform" on spotify, so I won't invest in having it there.
Hey, I run two somewhat popular music recommendation websites, the Music-Map and Gnoosic. I currently experiment with the ability to feature the music of artists, so I have put your music up on Gnoosic:
I'm not yet sure when I will turn the signup into a public feature. Currently, I'm interested in feedback from artists. If you like, shoot me an email. The two sites have a pretty big audience, and I would like to find out how I can use it to help small artists.
I love your music! Thanks for sharing your insight. I'm happy to say that verifying and posting from other services is coming... we just chose to start with Spotify. We are a small team of two working on this as a passion project so have to build little by little. We will keep you updated :)
Thank you! And I absolutely understand you gotta start somewhere.
Music recommendation is HARD. A lot of people don't think they like didgeridoo (which is where my expertise lies) until they hear a live performance of contemporary didge over a PA and they're in tears of joy (seriously, if you got the chance, give it a try - the stack of vibrations from this single, simple tube is amazing). IME, that wouldn't work for your average ear-buds or "smilie curve"-tuned over-ears. So, say, a friend convinces you to go to some concert with them - that would be a proper, healthy, sustainable music recommendation service. But of course that doesn't scale in any direction.
Best of luck with your project - I'll be checking back :-)
I'm not trying to tell you how to market your music, you do you, but with services like Tune Core, Distro Kid, etc that allow you to submit your music to every single music service in existence with one click, why wouldn't you just put your music everywhere? You may be surprised at just how big of a community there is for your niche genre on Spotify and elsewhere!
Anyways, I agree that the Spotify requirement is artificially limiting, especially if this is targeting up and coming artists. They should probably support all of the biggies (Spotify, Apple Music, Youtube, etc) as well as indies like Sound Cloud, Band Camp, etc.
I've pondered on and off about distrokid (or the like), and thankfully one of my peers has done just that. Mind you, he is more established and "successful" (as in, gets more paying gigs, more audience, ...). So I've talked to him about it. During the past half year, he's "earned" 4€ from Spotify (in other words, one song made the cut of 1k listens over a year), and 180€ from Bandcamp.
I see Distrokid as helpful marketing investment for discovery, so I was still inclined to burn some money (one teaching session will likely cover my loss, so whatever), but then he went on how he put a video of a performance of a published song of his (published via Distrokid) on YT - et voila - he's got challenged by Distrokid for it. Uh, yeah. I understand you gotta relinquish rights to let DK do their thing, but that's the sort of headache I wouldn't ever want. Maybe he was doing something wrong, of course, but I'll likely fall into the same trap. More reading required ...
I'll keep in touch with him about it, maybe the discovery on spotify led to the BC throughput, but we'll have to dig deeper. There's a lot of us small artists in different stages of our development (say, performance, entertainment, composition wise), I'd love to exchange experiences on what works and what doesn't for the ailing musician. There's platforms which contains a lot of us, but everybody seems to be struggling and there seems to be a strong sense of competition among many of my peers, so it's hard to get unfiltered (subjective) truths there. And the internet is filled with horror stories as well as "from dishwasher to millionaire" survivorship biased fairy tales. Oooph.
ALL music discovery algorithms are shit. The only thing good at recommending music to humans is other humans. But AI features are so much easier to build than social features, smh. Spotify's social features are worse than abysmal.
The implicit assumption here is that popularity is a good heuristic for relevance. Typically music recommendation systems will take popularity into consideration, but it is not the primary ranking factor.
Every day people order the top 5 songs and nominate one another song and a new topic. That way, we always have songs on the same topic. When there are more than 5 songs nominated on the same topic - people filter them out with one extra step.
It's fun, and it's been running since Nov 2022 https://urbanpoll.com/dj/rounds. I'd be happy if more people joined. I'm the "founder" of this little side project.
Amazing, history does repeat itself. One of the companies in my batch (S07) was iJigg (https://www.crunchbase.com/organization/ijigg), which had the 2007's version of this interface: single upvote, player, and all. It did well and then petered out as many hyper-niche apps did at that time.
There is also https://hypem.com/, which focuses on music posted by blogs but is otherwise fairly similar. It helped me discover a lot of cool music back in the day :) Nowadays Spotify's music discovery algorithm is often "good enough", though.
If we think about the technology available around 2007, a crowd-sourced music recommendation system is basically what I would expect. No offense, that’s just what we had at the time. The Netflix prize wouldn’t come out until 2008. Pandora’s Music Genome project was perhaps ahead of its time but mostly manual (20-30 minutes per song!).
Today, we would usually use automatic song similarity technologies to recommend songs. The reason is pretty clear: popularity-based systems don’t know anything about the songs they recommend. They don’t know anything about you. They only know metadata about the song, that a certain number of people liked a song, essentially throwing away half of the information in a user-song tuple.
While a nice idea, there's just so much more music made than SaaS products and this will eventually need to turn into a curated list ala spotify/apple music to keep users from drowning in static.
I think highlighting a specific type of music and finding your niche audience will be the best way towards a happy userbase.
I've been predicting a return to manual content curation for a while. Algos are not going to be able to keep up with the flood of new content. Record labels are going to be more important than ever.
I’m big into Doom/stoner rock/space rock/etc, so I subscribe to Weedian which has a fire hose of download codes for new releases/artists.
Even for this fairly niche group of genres, the free codes I get have filled my bandcamp artist list to the point of it no longer being useable for its intended purpose.
It was already getting unwieldy with just my classicly paid purchases (which I still do a lot of), I have to resort to manually create artist playlists to navigate now.
Even just the Weedian compilation albums are so full of tracks, I just don’t have the time to listen through them to evaluate and brutally curate which songs/artists I love before the next one comes out. The lasted album, “Trip to California” is like 99 songs and over 8 hours. I think people vastly underestimate how much music is out there and is being newly released.
Part of this, is the Doom genre tends to be 5-15 minute songs, which is absolutely what I want, but also makes listening a significant time investment. Fortunately the majority of songs are easy to say “yeah nope” to and cull.
Definitely suffering from success here. On the plus side, I’ve come across a lot of gems that can deserve to sit next to Yob on my playlist, that I wouldn’t have come across otherwise.
I really need figure out solution that lets me stream from a self-hosted archive, rate songs on a 5 star scale, genre tag, and share playlists with family members. The 5 star rating seems to most troublesome to fulfill.
A tool to automatically produce a public facing playlist that links to bandcamp, YouTube and perhaps other places as a last resort would also be amazing. The best way to discover music, is finding and following people like me who make exploring and curation into a hobby. “What music do I like? Here’s my 5 star list”
Couldn't agree more. We are actually already saving genre against any track submission so when we have a big enough bank of music we will switch on the genre filters :)
What is the best Pandora-adjacent service these days? In other words, where can I plug in "Artist X" and get a great platter of recommendations?
Spotify doesn't cut it for me -- the recommendations are either hyper-generic or songs that I've listened to many times in the past. Last.fm doesn't seem to have many lights on these days.
Pandora, in its early days, was a great place to discover new music because it would find songs that were roughly similar to what you requested but by often unheard-of artists.
I use a site called https://www.queup.net which is a successor to turntable.fm. You find a "room" with a genre of music you like and just sit back and listen to people DJ using soundcloud and youtube videos. I have it on all day in a room that is majority electronic/chill/lounge and mostly non-vocal. I've learned about so many artists this way but you need to find an active room you like, and sites like these tend to skew towards the many different electronic subgenres
I haven't personally used it, but I do know that ListenBrainz offers a Last.FM-esque experience, and I see it does have a "similar artists" section. I'm in a similar boat as you though. Most of the Spotify-generated stuff is just repeats of songs I've already liked or just generic songs.
I've been diving into the past about artists I've been meaning to check out and also using Rate Your Music to stay semi-conscious of what's popular today.
Bandcamp has a "if you like this, you may also like" feature which seems to be unbiased by listening history and allows unheard-of artists to crop up. I've found a few good recs that way although it can be hit-or-miss.
I used to use Last FM for this, but no longer. I think this needs to be brought back on modern services. An artist might be in a totally different genre but if they have the same ‘feel’ you might like both
Out of all the ways discovering music, surprisingly Soundcloud has been the best. Being able to curate a list of DJs in various genres means that I'm constantly finding new music by being exposed to what they're putting in their sets.
It's kind of a meta music discovery, I can piggy back off others who have tastes that I like.
The biggest benefit of this is that I don't necessarily get recommendations of music that's adjacent to what I already like. In other words, if I like Depeche Mode, I don't want to listen to bands that sound vaguely like them.
What I've yet to see someone do, is track what music someone listens to after listening to a song. Often when I hear a piece of music, it makes me want to listen to something else that feels the same way or reminds me of being in similar place in time. A music recommendation engine based of this concept would be interesting.
Love this, I totally agree. That's why we want to find a way to make this work without algorithms. We know what hits us in the same way as something else. I'd love to build this site to achieve close to what you suggest.
Doesn't this work better in the context of personalized recommendations? Pandora is scarily good at this because it's an offshoot of the Music Genome Project, which analyzes the constituent parts of music (https://en.wikipedia.org/wiki/Music_Genome_Project?wprov=sfl... )
It combines that with your personal things up and listening history and predicts what other songs you yourself are likely to enjoy.
By contrast, up votes from strangers across genres seems unlikely to match your personal tastes.
I don't think you'll be able to compete with Spotify for personal recommendations. But there's a fatigue that comes with being kept in your recommendation bubble. I want to hear something new and good and different! And not corporate-approved like Pitchfork (rip). Through the grace of god Hype Machine is still running and still highlighting some solid tracks.
Strangers' preferences can be better than a personalized recommendation, if they are cool strangers.
Last edit: I seem to remember a study that concluded that people who listened to a wider range of music types spent more money, or were more lucrative advertising targets. However, there are fewer of those kinds of listeners.
Maybe because by "wider range" many people understand whatever the trendy stations will feed them - from rap to dad rock, r&b or pop, so they behave as a captive audience? My Spotify must have a reason for always trying to feed me Coldplay although I swear I never (actively) listened to anything from them.
Genre filters are coming, once we have enough tracks live. What we want to do it give artists and music lovers a place to discover talent outside of what might be served up for them by algos. If the community likes a track enough it will rise to the top and get in-front of more ears that might not have discovered it otherwise.
I tried Pandora once, it just played me music that was a bit like the music I liked, but worse. If the music were better I'd already have known about it
I built something just like this, but I found everyone would drop there tracks and never look at anybody else's stuff. I shut it down as it ended up being kidn of pointless.
It doesn't make sense because artistic tastes are idiosyncratic, making popularity as the sole criterion worthless, and the volume of art work overwhelms anyone's ability to even superficially consider.
The only scalable solution is to use machine learning, as all the big players do. They're never going to be much better than Spotify. I'd be surprised if they got even close.
I suppose there might be a use case for A&R types who are on the hunt for new material, rather than casual listeners.
Good points here. What we want to achieve is something akin to the 'support' act before a headline set. Where a more successful artist champions a smaller artist because they like what they are doing. With this site the goal is for artists to share fans to help grow their fan base. This version is the first step towards that. If that makes any sense :)
Readit News
E.g., here's some unexpected percussive organic trance for you that my collaborators and I put on BC: https://dojorich.bandcamp.com/album/tubecolab (I'm Freddie Veggie, btw). Music of that scope doesn't "perform" on spotify, so I won't invest in having it there.
https://www.gnoosic.com/artist/freddie+veggie
And put you on the map next to some other didgeridoo artists:
https://www.music-map.com/freddie+veggie
I'm not yet sure when I will turn the signup into a public feature. Currently, I'm interested in feedback from artists. If you like, shoot me an email. The two sites have a pretty big audience, and I would like to find out how I can use it to help small artists.
Music recommendation is HARD. A lot of people don't think they like didgeridoo (which is where my expertise lies) until they hear a live performance of contemporary didge over a PA and they're in tears of joy (seriously, if you got the chance, give it a try - the stack of vibrations from this single, simple tube is amazing). IME, that wouldn't work for your average ear-buds or "smilie curve"-tuned over-ears. So, say, a friend convinces you to go to some concert with them - that would be a proper, healthy, sustainable music recommendation service. But of course that doesn't scale in any direction.
Best of luck with your project - I'll be checking back :-)
https://open.spotify.com/search/didgeridoo/playlists
Anyways, I agree that the Spotify requirement is artificially limiting, especially if this is targeting up and coming artists. They should probably support all of the biggies (Spotify, Apple Music, Youtube, etc) as well as indies like Sound Cloud, Band Camp, etc.
I've pondered on and off about distrokid (or the like), and thankfully one of my peers has done just that. Mind you, he is more established and "successful" (as in, gets more paying gigs, more audience, ...). So I've talked to him about it. During the past half year, he's "earned" 4€ from Spotify (in other words, one song made the cut of 1k listens over a year), and 180€ from Bandcamp.
I see Distrokid as helpful marketing investment for discovery, so I was still inclined to burn some money (one teaching session will likely cover my loss, so whatever), but then he went on how he put a video of a performance of a published song of his (published via Distrokid) on YT - et voila - he's got challenged by Distrokid for it. Uh, yeah. I understand you gotta relinquish rights to let DK do their thing, but that's the sort of headache I wouldn't ever want. Maybe he was doing something wrong, of course, but I'll likely fall into the same trap. More reading required ...
I'll keep in touch with him about it, maybe the discovery on spotify led to the BC throughput, but we'll have to dig deeper. There's a lot of us small artists in different stages of our development (say, performance, entertainment, composition wise), I'd love to exchange experiences on what works and what doesn't for the ailing musician. There's platforms which contains a lot of us, but everybody seems to be struggling and there seems to be a strong sense of competition among many of my peers, so it's hard to get unfiltered (subjective) truths there. And the internet is filled with horror stories as well as "from dishwasher to millionaire" survivorship biased fairy tales. Oooph.
Every day people order the top 5 songs and nominate one another song and a new topic. That way, we always have songs on the same topic. When there are more than 5 songs nominated on the same topic - people filter them out with one extra step.
It's fun, and it's been running since Nov 2022 https://urbanpoll.com/dj/rounds. I'd be happy if more people joined. I'm the "founder" of this little side project.
Is the statement not valid for products too ?
Best of luck with this version!
Today, we would usually use automatic song similarity technologies to recommend songs. The reason is pretty clear: popularity-based systems don’t know anything about the songs they recommend. They don’t know anything about you. They only know metadata about the song, that a certain number of people liked a song, essentially throwing away half of the information in a user-song tuple.
I think highlighting a specific type of music and finding your niche audience will be the best way towards a happy userbase.
Even for this fairly niche group of genres, the free codes I get have filled my bandcamp artist list to the point of it no longer being useable for its intended purpose.
It was already getting unwieldy with just my classicly paid purchases (which I still do a lot of), I have to resort to manually create artist playlists to navigate now.
Even just the Weedian compilation albums are so full of tracks, I just don’t have the time to listen through them to evaluate and brutally curate which songs/artists I love before the next one comes out. The lasted album, “Trip to California” is like 99 songs and over 8 hours. I think people vastly underestimate how much music is out there and is being newly released.
Part of this, is the Doom genre tends to be 5-15 minute songs, which is absolutely what I want, but also makes listening a significant time investment. Fortunately the majority of songs are easy to say “yeah nope” to and cull.
Definitely suffering from success here. On the plus side, I’ve come across a lot of gems that can deserve to sit next to Yob on my playlist, that I wouldn’t have come across otherwise.
I really need figure out solution that lets me stream from a self-hosted archive, rate songs on a 5 star scale, genre tag, and share playlists with family members. The 5 star rating seems to most troublesome to fulfill.
A tool to automatically produce a public facing playlist that links to bandcamp, YouTube and perhaps other places as a last resort would also be amazing. The best way to discover music, is finding and following people like me who make exploring and curation into a hobby. “What music do I like? Here’s my 5 star list”
What is the best Pandora-adjacent service these days? In other words, where can I plug in "Artist X" and get a great platter of recommendations?
Spotify doesn't cut it for me -- the recommendations are either hyper-generic or songs that I've listened to many times in the past. Last.fm doesn't seem to have many lights on these days.
Pandora, in its early days, was a great place to discover new music because it would find songs that were roughly similar to what you requested but by often unheard-of artists.
This is made by the MetaBrainz folks, the same people behind the MusicBrainz database (which is used by ListenBrainz).
I've been diving into the past about artists I've been meaning to check out and also using Rate Your Music to stay semi-conscious of what's popular today.
It's kind of a meta music discovery, I can piggy back off others who have tastes that I like.
The biggest benefit of this is that I don't necessarily get recommendations of music that's adjacent to what I already like. In other words, if I like Depeche Mode, I don't want to listen to bands that sound vaguely like them.
What I've yet to see someone do, is track what music someone listens to after listening to a song. Often when I hear a piece of music, it makes me want to listen to something else that feels the same way or reminds me of being in similar place in time. A music recommendation engine based of this concept would be interesting.
It combines that with your personal things up and listening history and predicts what other songs you yourself are likely to enjoy.
By contrast, up votes from strangers across genres seems unlikely to match your personal tastes.
Strangers' preferences can be better than a personalized recommendation, if they are cool strangers.
Last edit: I seem to remember a study that concluded that people who listened to a wider range of music types spent more money, or were more lucrative advertising targets. However, there are fewer of those kinds of listeners.
The only scalable solution is to use machine learning, as all the big players do. They're never going to be much better than Spotify. I'd be surprised if they got even close.
I suppose there might be a use case for A&R types who are on the hunt for new material, rather than casual listeners.