Assigning moods to tracks

[Bernhard Bliem]

It could be useful to assign a mood to a track by assigning numerical values to categories like elation (how melancholic / euphoric it is), intensity (how well you can sleep while it's playing ;)) and dedication (how much you have to concentrate in order to be able to enjoy the track). Then Exaile could populate the playlist with tracks that suit the user's current mood by choosing tracks which correspond to his given values.

[reacocard (reacocard)]

An interesting idea, though ideally we would have some way of scanning the file to automatically determine the mood (there are a number of libraries which attempt to allow this sort of analysis).

[Veloce (kerry-mayes)]

I'd been thinking about this independently this week. I've done a little research around the web. There is a group working on algorithms for Audio Music Mood Classification. They seem to have an annual algorithm competition.

http://www.music-ir.org/mirex/2009/index.php/Audio_Music_Mood_Classification

They have obviously done some respectable work on classifying songs into five "mood clusters".

There is also a chap writing something in Python atm:

http://olug.cs.oberlin.edu/~serickso/about.html

However he is working independently so has his own ideas on mood classification.