Pandora Radio has a pretty damn cool and exclusive feature that no other music streaming service has, the Music Genome Project:
I realize it’s a lot of work to to develop such functionality, however, given that the community of ListenBrainz is relatively large, it should be possible to implement this here and let us, the users, tag and edit songs with “music genes” and describe what sort of accoustics and instruments are involved in each song, and then a few ListenBrainz developers could work on the music genetics algorithms that match similar songs. Last.fm already has a similar algorithm, so that part is not exclusive for Pandora’s Music Genome Project, however, what makes the MGP stand out is the unique genetics of each song, which is really cool and nerdy.
On Pandora Radio, if you look up various songs, you will see descriptions like this (this is taken from Bonnie Tyler - Total Eclipse of the Heart):
FEATURES OF THIS SONG
Pop rock qualities Mild rhythmic syncopation Acoustic rhythm piano Repetitive melodic phrasing Interweaving vocal harmony
This is part of the “genetics” of each song. Other songs have different “features”, but this is really cool and useful in any case, for organizing music.
The problem with Pandora’s MGP is that it’s proprietary, closed source and a trade secret (the music genes involved are not public), and therefore we really need an open source equivalent to the Music Genome Project. And given that MetaBrainz is non-profit and pro-open source, I think ListenBrainz is the perfect website target to implement the development of an open source clone of the Music Genome Project. This would really make ListenBrainz so much better, if we had an open source MGP-like clone to tinker around with.
Yeah, AcousticBrainz is definitely similar to the Music Genome Project. However, with the MGP, each song is actually analyzed by professional musicians. That’s why I suggested that the ListenBrainz user community could do the editing of tags, genres, composers and other details and descriptions that aren’t always picked up by audio software. It would also make ListenBrainz a lot more fun if it had an open source equivalent to the Music Genome Project. Most users on ListenBrainz never check AcousticBrainz anyway (and AcousticBrainz entries are basically hidden among all the details in MusicBrainz entries).
I realize audio software like Picard and other tools, are a lot more accurate in determining various features of each song, than ordinary laymen like me who don’t know how to compose songs and so on, but still, user input could be valuable too. As I was saying, Pandora Radio employs actual professional musicians for analyzing the “genes” (attributes) of all songs, and I guess the same can be done here.