Information Technology Reference
In-Depth Information
Figure 7. Musical Genre Principal Components Analysis.
cross-reference the ID number to the song, where
appropriate. We feel that this small selection of
songs represents a reasonable cross-section of
contemporary popular musical genres.
Our main motivation in this area of research
and development was to address some of the
shortcomings traditionally employed in automatic
recommendation and playlist generation tools.
Historically, these tools evolved in a similar
way to that of Automated Collaborative Filters
(ACFs). That is to say, simple measurements of
user preference and the preference of a typical
population were used to build a ranked table of
music in a library. These analysed information
such as the most frequently played tracks, a user
rating of each track, favourite artists and musical
genres, and other meta-data attached to a song
(Cunningham, Bergen, & Grout, 2006). However,
this is not to totally trivialise the area of automatic
playlist generation, since a number of systems exist
that employ much more advanced learning and
responsive Automated Music
Playlists
Some of our most recent and cognate work com-
bining the use of emotion, content and context
in musical applications has been in the area of
intelligent playlist generation tools and this work
is explored in greater detail in a separate work
(Cunningham, Caulder, & Grout, 2008). However,
to see the effectiveness of combining all three of
these areas, the reader is provided here with a
summary of that work to date.
Search WWH ::
Custom Search