Boffins in California say they have produced music-recommendation software which produces playlists "as good as" those from Apple's iTunes Genius - and which has the advantages of collecting no user data and having in its repertoire a lot of music that "Genius knows nothing about".
An eccentric Genius.
Details on two different experimental systems were presented by UC San Diego engineers today at a music-info conference in Kobe, Japan.
"Our goal is to make a music recommendation tool that is as good as or better than Genius, but that does not require massive amounts of user data," said Luke Barrington, PhD student at UCSD. "The system we are developing can analyse and recommend completely unknown songs by new bands as accurately as it analyses the most popular hits."
Barrington and his collaborators believe that automated media search will become critically necessary in a world where millions of different songs or videos can be accessed via online services.
"Playlists are an increasingly important tool for overcoming what otherwise might be an overwhelming amount of streaming music," said Barrington.
However he and his colleagues consider that Apple's Genius isn't smart enough, and in any case can't recommend music which isn't on iTunes. Furthermore it appears to make recommendations by monitoring the habits of its users, rather than by analysing the music. At least one disgruntled user tells the Reg that this can produce some bizarre recommendations (see pic), and Barrington and his chums believe that Genius naturally tends to offer a user established bands rather than unknown ones that they might in fact prefer.
"We don't think Genius actually knows anything about the acoustics of songs," says Barrington. "The data it is using is built upon acoustic analyses of the music done by millions of humans."
Not so Barrington and Co's experimental recommender-ware, which apparently already produces playlists rated by humans as "as good as" Genius, and is expected to do better.
"Our computer system works by listening to the music – it doesn't know anything about artists or albums or charts... We weren't expecting our system to beat Genius at making playlists based on the most popular songs – our system doesn't know about artists, popularity, release dates, albums or anything else that the average music fan is aware of. Once we add that information in, we think we can build something that is really smarter than Genius," said Barrington.
Barrington and his fellow music-engineers' paper, Smarter than Genius? can be read in pdf here. ®