Interestingly I found a lot of new music that way and also found other music directions that suddenly interested me.
Thinking this a little further, it provides the perfekt basis for a dating application. Music is a very personally representation of oneself and if you find someone who has the same muscial taste, you already have a common basis for communication.
Obviously this never made it past developer status since Spotify does not like third party apps storing their data (i.e. playlists of other people). You can use the data but not store it or analyse it.
It made me think that Spotify should build a dating app since they would have no interest in keeping you in their dating app as thats not their main business, unlike other dating platforms that have little interest in losing their customers.
I don't think a shared taste in music is that important, or at all, in a relationship but this is the premise of aptly named Tastebuds[1], a dating app based around favorite music. You can import your listening data from Last.fm and Spotify.
Spotify doesn't have its claws in real-life gatherings yet, but they might have something with their music (personality) and podcasts (interests, affiliations) data.
So, no music taste isn't ultimate be-all-and-end-all factor but for some it can be very important!
Tastebuds seems only to have a facebook login - hm, not everyone uses FB! Strange that they don't have a Spotify login, probably the same issues that other apps face when they start to store Spotify data.
I think partly it's just familiarity. There are plenty of "classics" I didn't really like when they came out but now they bring a feeling of nostalgia so even though they weren't "me" back then, for some reason they now feel like part of me, my history, my experience.
I have a largish group of friends that I met through similar tastes in music and being "neighbours" on last.fm.
Slightly off-topic, but originally I had planned to organically grow the playlist count by allowing users to login with their Spotify account and reading their playlists. I even made a prototype...but turns out Spotify only allows working with user data once you have your API quota extended, which seemed like quite the undertaking.
I even went as far as setting a second developer account to have a larger limit ;)
I also did a writeup[1] on lessons learnt in case someone is interested in rinsing and repeating on this idea.
[0] true story, multiple times
Personally I'd get far more value if I could chuck in 3-4 favourite songs and see which playlists contains all of them (or order by greatest number of matches first); this would allow me to more accurately match my tastes to discover a potentially new bunch of other songs I might like.
If there is a way to match against more songs, apologies! The post title implies there is, but the site seems to suggest it's limited to one song.
I really want to see if anyone else has Left Hand Free and Pop Muzik on a playlist together, and if they've found any _other_ songs that contain eeny-meeny-miney-moe in the lyrics and made a playlist of them like I have...
(Here's another oddball one, if you're a puzzle type. Bring The Magic - Steve Wynn. So It Goes - Nick Lowe. Roadrunner Once - Jonathan Richman. Heavy Metal - Sammy Hagar.)
It is currently limited to one song, but there is already an issue on github planning multiple song selection. Just haven't gotten around to it yet.
As for over-abundance of results, sorting by name length in ascending order should help at least somewhat, but the search approach is an open problem.
As an aside: on large screens there is also an option to choose a specific song by dragging it from the Spotify client.
https://www.google.com/search?q=%22state+of+grace%22+%22I%27...
i.e. in practice, for practical metrics, it probably takes a lot (I'm not going to guess how many) more.
First track couldn't be found. Second track could be found, but it couldn't find the aforementioned playlist (with almost 200k likes). Third song also can't seem to be found.
So I'm thinking that the indexing is not complete. Unless I misunderstand something.
Currently the database contains only ~18k playlists and ~1.4M tracks. So only the relationships between them are used in the app.
Edit: theres even a counter at the top of the page, but it takes a while to show up, because of the HN effect.
Oh and a bug: If I first search for something and get a result, then enter a new search and scroll down before this new search has completed, it starts searching for page 2-7 of this new search term. Even better if you've already triggered the search for page 2-7 on that first term. If you then scroll down after updating the search, it is now searching for page 8-11 of that new search term :)
The loading indicator is also on the list, TBH was supposed to go up before ShowHN, but I forgot. And yeah, now the query execution wait time for my measly 2-core postgres instance is about 30 seconds :D.
It's kind of weird that I only found this song in your playlists, and not in one of my playlists which also contains this song. For example: https://open.spotify.com/playlist/0IntLgAqI2EysEvqwiivXQ?si=...
Anyway, I'm checking out your "On Repeat" playlist, however it might be better if your personally generated playlists are excluded? I'm not sure.
Really cool service!
The playlist in case you want to use it for testing later: https://open.spotify.com/playlist/5DcbwcCkBde68RgsxO2Sn2
Ideas for improvement:
* autocomplete
* autocomplete with song, artist and album image
* allow for search by track & artist name
* match up one of my playlists with a playlist you have in your DB
I maintain an archive of Spotify playlists, scraped daily via GitHub Actions: https://github.com/mackorone/spotify-playlist-archive.
If you're willing to share, I'd love to hear about your experience with the Spotify API. What endpoints do you use? How often do you call them? How do you handle rate limiting? Has Spotify reached out regarding storing playlist data (being in violation of their terms of service)? Any other lessons or advice?
Additionally, I wonder if there's any opportunity for collaboration here. Perhaps you can use the archive to fill out your dataset? Maybe it can act as a cache for your system's scraping needs? Perhaps our Web UIs could link to each other or be combined somehow (https://spotifyplaylistarchive.com/)?
> Currently we keep track of 18378 Spotify playlists
How did you find all of these playlists? Querying Spotify for "Spotify"-owned playlists gives me less than 5000 results. I'd love to add more playlists to my archive but I don't have good discovery mechanisms, especially for non-Spotify-owned playlists.
The bottleneck here is using the Spotify API to find relevant playlists - as others have noted here, Spotify's API doesn't provide a way to perform direct lookup of playlists containing a given track, so the best approach I've found is to perform many text-based searches for desired search terms (e.g. artist, song name), and then do breadth-first-search on playlists created by the same users from the initial result set, in order to find other playlists that might have the artist or song you are looking for.
[1] Available in very limited fashion at https://vybe.link; this is a Spotify beta application so the full app only works for whitelisted Spotify accounts
https://playlists.dags.dev/tracks/7xdkou0YaNLW9hWXU9MRcR/pla...
Tried a bunch more without success. Granted, I like a lot of obscure electronica.
Love this concept, way quicker than using the “found on” in the Spotify app - which doesn’t even always work that well.
Works great, thank you!
> Found 32646 tracks.
Yikes. Adding the artist to the search yields 0 results, and since it only loads a number of results until you scroll to the bottom of the page, I could not find the track I wanted.
It would be great to have the option to specify the author somehow
Mostly acoustic and piano based, lesser-known songs. Please add it to your index.
https://youtube.com/playlist?list=PLhMrEC5L8T4FmbQP78B5gzEJh...
Something useful I found is the "Go to song radio" in Spotify, that creates kind of a "dynamic" playlist with similar songs.
I also use the Spotify radios a lot, but they are heavily curated - they contain a lot of songs that you already have in a playlist or have liked. So they are useful for listening, but slightly less for strictly music discovery.
Disclaimer: I've built this.
Obviously this could be a lot better.
Vibin to a new playlist now.
I Would love something like this for Youtube.
