Find another service like me
Cuz I ain't seen one - Tyler, The Creator
I want Kapell to have really interesting capabilities. What about finding all the songs recorded at Electric Lady Studios in the 90’s by artists who’s names start with the latter J or top-charting songs for artists who’s birthdays are this month? That’s the dream!
To help realize this dream Kapell’s API is migrating to GraphQL. Musicbrainz is linked to Wikipedia by way of Wikidata, which (until now) was resolved at query/enrichment time. This added latency but was offset by L2 caching. To reach our goals we needed to bring that data into Kapell’s database. That work is now complete, lowering latency and improving enrichment hit rates.
But there’s more.
We want our future GraphQL work to be speedy and rich. To accomplish this we can “materialize” the underlying data which is rife with joins and rules. First, we can exclude any songs or albums that aren’t in Apple’s streaming catalog since Kapell is an Apple Music client. We’ll also go farther than Apple’s catalog by including information on the other artists credited in music: engineers, producers, etc. While we’re here let’s also grab information like recording and mastering studios.
Credits
666,081 artists appear in Kapell’s credit edges, meaning they contributed to the music. Of those only 111,760 or ~10% have an entry in Apple’s database. With this effort, Kapell gains the ability to relate those artists to their work.
Songs
27% (~5.3M) of the song IDs in Apple’s catalog are duplicate pointers to existing recordings: editions, deluxes, regional variants, etc.
Other Things
Over 300k record labels and more than 32,000 places are available for linking. This being openly available, volunteer-led data there are also gaps. Almost 420k artists have no credit links, leaving them informationally stranded.
Final Thoughts
Over 30 million edges now connect Kapell’s datasets — about six per node, counting artists, albums, labels, places, and series as nodes. Songs live on the edges themselves. This creates some really fascinating opportunities for deep, focused listening.
This doesn’t buy us a bigger catalog — we’re limited to Apple’s after all — but to bigger questions. What else came out of this room? What else did this person touch? What is this sample?
My goal, soon, is to listen to a single recording engineer. Not a rapper or a genre or an algorithmic playlist but a back-of-the-record, small-type artist who’s name is on music I love, and who otherwise I’d have no way to follow.
Happy enriching!