aboutsavedsign in
×
scenes by Lux Cache is a music discovery tool. you give it people or releases you trust, it crawls public data to find what else those networks are into.
eight modes — soundcloud, bandcamp, discogs, rateyourmusic, nina, resident advisor, nts, and music blogs. each works the same way: seed with taste you trust, scrape the network around it, surface overlap.
if multiple independent sources all point to the same thing, it shows up. that's it. no algorithms, no machine learning.
×
saved
×
Sign in with Lux Cache
or
Don't have a Lux Cache account? Sign up at luxcache.com
scenes uses your Lux Cache account for saving and syncing data.
+
1. add ears — soundcloud profiles whose likes you trust. curators, DJs, diggers.
2. hit scrape — pulls their public likes via the API. takes a minute or two.
3. consensus view shows artists liked by multiple ears. if 3 ears independently liked the same producer, that's a signal.
4. all likes shows the raw feed, most recent first. no overlap filter.
5. save groups of ears as scenes to switch between them.
ears
depth
recent likes. faster scrape, good for a quick scan.
scenes
+ save current ears as scene
+ soundcloud client_id (optional)
open soundcloud, press F12, network tab, play any track, find client_id= in any request URL.
1. add seeds — bandcamp artists or releases you love. names or full URLs.
2. hit scrape — finds fans who bought your seeds, then crawls their public collections.
3. results rank releases by how many of those fans also bought them.
4. save groups of seeds as branches. more seeds = broader net, better signal.
seeds
depth
fewer fans sampled. faster scrape, broader signal.
branches
+ save current seeds as branch
1. add diggers — discogs usernames whose collections you trust.
2. paste your discogs token (settings > developers > generate token).
3. hit scrape — pulls each digger's public collection and wantlist via the API.
4. results rank releases by overlap. have + want split visible. everyone wanting but nobody having = strong signal.
5. save groups of diggers as crates.
api token
required. 60 requests/min free.
diggers
crates
+ save current diggers as crate
1. add seeds — rateyourmusic album URLs you love.
2. hit scrape — finds users who rated your seeds highly, then pulls their rating histories.
3. highest rated ranks by avg rating from your taste-matched cohort. most found ranks by pure overlap count.
4. save groups of seeds as lists. scraping is intentionally slow (~2s between requests).
seeds
depth
15
raters per seed. higher = richer results, much slower.
lists
+ save current seeds as list
1. add seeds — nina releases or artists. paste URLs from ninaprotocol.com.
2. hit scrape — finds wallets that collected your seeds, then pulls their other purchases.
3. results rank releases by collector overlap. all data is on-chain and public.
4. save groups of seeds as sets.
seeds
depth
20
collectors per seed.
sets
+ save current seeds as set
1. add seeds — artists you rate. names or resident advisor URLs.
2. hit scrape — searches RA for events they played, pulls full lineups.
3. results rank artists by booking overlap. same promoters trusting the same artists is the signal.
4. save groups of seeds as circuits.
seeds
depth
10
events per seed. higher = wider booking network, slower.
circuits
+ save current seeds as circuit
1. add seeds — artists you love, or NTS show URLs directly.
2. hit scrape — finds shows that played your seeds, pulls full tracklists.
3. most shows ranks by raw appearance count. most connected ranks by how many seeds they co-appeared with.
4. save groups of seeds as mixes. every tracklist pick was an active curatorial choice.
seeds
depth
10
episodes to pull per seed. higher = deeper, slower.
mixes
+ save current seeds as mix
1. add sources — music blog homepage URLs. tries RSS first, then scrapes HTML.
2. hit scrape — pulls recent articles, extracts bandcamp, soundcloud, and other outbound links.
3. most covered ranks by raw source count. most recent prioritises newer mentions.
4. save groups of sources as feeds. embedded players are structured data — stronger signal than text.
sources
depth
10
articles per source.
feeds
+ save current sources as feed
reset all data