Digital Signal Processing Reference
In-Depth Information
Chapter 11
Applications in Intelligent Music Analysis
Of all noises, I think music is the least disagreeable.
—
Samuel Johnson
As digitised music has conquered the market for more than ten years, advanced
techniques of MIR are gaining interest and importance. Caused by the progress in
lossy perceptual audio coding (MP3 and similar), broadband internet connections and
high volume storage capacities, large music databases have emerged which demand
novel handling strategies [
1
-
3
].
The increasing popularity of portable music players and music distribution over
the internet has made worldwide, instantaneous access to rapidly growing music
archives possible. Such archives must be well structured and sorted in order to be user
friendly. For example, many users face the problem of having heard a song they would
like to buy but not knowing its bibliographic data, i.e., title and artist, which is nec-
essary to find the song in conventional (on-line) music stores. According to Downie
in [
4
], almost three fourths of all MIR queries are of bibliographic nature. The query-
ing person gives information he or she knows about the song, most likely genre, metre,
tempo, lyrics or acoustic properties, e.g., tonality, and requires information about title
and/or artist. In order to have machines assist in building a song database queryable
by attributes such as tempo, metre or genre, intelligent Information Retrieval algo-
rithms are necessary to automatically extract such high level features from raw music
data. Hence, many new tasks in private as well as in professional environments have
occurred, as for example rhythm recognition [
5
,
6
], genre [
7
] and mood classifi-
cation [
8
], melody extraction [
9
], chord detection [
10
] or key determination [
11
].
Many works exist that describe or give overviews over basic MIR methods, e.g.,
[
5
,
12
-
20
].
In this chapter, an overview shall be given on intelligent music analysis. The
selected application scenarios cover methods on rhythmic aspects first: The sep-
aration of drum-beats [
21
] is followed by the determination of onsets in music
[
22
,
23
], and tempo, metre and ballroom dance style determination [
6
,
24
-
27
]. Sub-
sequently, analysis of the tonal aspects, reaching from musical key [
28
] to chords
