Music

was reminded of Khruangbin's incredibly cool playlist website, 'Shelter in Space', where you choose the activity you're doing and how long you want to do it for, and a playlist will be generated from an excellent pool of songs curated by Khruangbin

was reminded of Khruangbin's incredibly cool playlist website, 'Shelter in Space', where you choose the activity you're doing and how long you want to do it for, and a playlist will be generated from an excellent pool of songs curated by Khruangbin 

https://space.airkhruang.com/

I found an article by Lee Martin, the developer, which details the elements used to assess and categorise songs through Spotify's audio features: acousticness, danceability, energy, instrumentalness, liveness, loudness, speechiness, valence, and tempo. 

The algorithm then matches these features to the best activity.

I've always made playlists based on vibe and feel — e.g. similar sounds, juxtaposition, surprising combos, lyrics, genres, time periods, voices, instruments — so to see a method explicitly laid out like this and to think of songs for 'reading' as requiring high instrumentalness and low speechiness is fascinating

full article here, which also explains the changes in approach from an earlier playlist project called AirKhruang — the music for that was excellent too, and you can still find saved playlists from it on Khruangbin's artist playlists on Spotify

full article here, which also explains the changes in approach from an earlier playlist project called AirKhruang — the music for that was excellent too, and you can still find saved playlists from it on Khruangbin's artist playlists on Spotify

https://leemartin.dev/sheltering-in-space-with-khruangbin-f185885089e6