Digital Signal Processing Reference
In-Depth Information
Table 11.20
Chord types and emotions associated with these [
156
]
Chord type
Example
Associated emotions
Major
B
Happiness, cheerfulness, confidence, satisfaction, brightness
Minor
Bm
Sadness, darkness, sullenness, apprehension,
melancholy, depression, mystery
Suspended fourth
Bsus4
Delightful tension
B
7
Major seventh
Funkiness, moderate edginess, soulfulness
Bm
7
Minor seventh
Mellowness, moodiness, jazziness
B
maj
7
Major Major seventh
Romance, softness, jazziness, serenity, exhilaration, tranquillity
B
9
Ninth
Openness, optimism
Diminished
Bdim
Fear, shock, spookiness, suspense
B
7
/
9
Seventh, Minor ninth
Creepiness, ominousness, fear, darkness
B
add
9
Added ninth
Steeliness, austerity
Table 11.21
Recognised
Chord type
Example
chord types
D
Major
Minor
Em
C
7
Major seventh
Am
7
Minor seventh
F
maj
7
Major Major seventh
C
m
maj
7
Minor Major seventh
A
+
Augmented
Diminished
Fdim
Edim
7
Diminished seventh
For chord determination in the original music file, a fully automatic algorithm
[
157
] is used as was explained in Sect.
11.5
. It basically compares the chromagram
with predefined chord templates (cf. Sect.
11.4
), and outputs the chord type (e.g.,
major, minor, diminished) and the chord base tone (e.g., C, F, G
).
As chord features, 'bag-of-chords' is used with the frequency of occurrence of
a chord normalised to the total number of chords in a musical piece. Overall, 22
numeric features are obtained, of which the last simply is the number of recognised
chords (cf. Table
11.21
for those recognised).
11.7.1.3 Metadata
Rich meta-information for all music in the NTWICM database is hard to obtain given
which is available for each song. The year of release is used 'as is' as a numeric
feature. As for artist and title, by standard word delimiters text strings are chunked to
words. Then, the Porter stemming algorithm [
158
] is used and binary BoW features
