> Different idea but still related to time information. I was thinking > yesterday about maybe using time information to be able to recommend > music for your current mood. If we can do data mining to get stuff > like 'this song is usually listened soon before/after these other > songs', you could then just start with a song and let libre.fm > recommend songs that you or other people usually listen > around that one.
Last.fm exposes this data through the similar tracks and similar artists methods of the API. I have built a music player plugin that uses that data (and/or, optionally, local acoustic analysis) to queue up tracks that are similar to the one that is currently playing. The plugin was built for quod libet, but works with rhythmbox and mpd as well, and is built to be player independent, so should be easy to port to other players that support python plugins. You can find it here: http://code.google.com/p/autoqueue/ It would be awesome if libre.fm also provided similarity data, however one wishes to compute that, but I imagine it's computationally hard, easy to get wrong (google for freakommendations), and storage intensive, so it might not be low hanging fruit. > Another thing I don't like much about last.fm is the fact that you > only have one radio station mixing all the music you like. It would be > probably neat to use the tag information to group the music you like > by styles, giving a kind of personal radio stations by style. That way > you could do stuff like 'listen to my classical music radio station' > or 'listen to my Spanish music radio station'. I do this client side, again through said plugin, where I can restrict the stuff that is being queued through quod libet's awesome search engine. For instance I can queue similar tracks restricted to artists that will play at a festival I plan on going to. (Provided I have tagged my music by those artists accordingly, or have it in one specific place on the filesystem.) -- - eric casteleijn http://thisfred.blogspot.com _______________________________________________ Libre-fm mailing list [email protected] http://lists.autonomo.us/mailman/listinfo/libre-fm
