Information Technology Reference
In-Depth Information
(i.e. roughly an 80% return rate). For a non-researcher, it is quite dif
cult to apprehend
the scienti
c results in a both artistic and playful
way; scientists were also directly in front of the audience, which the audience
appreciated. All those factors made the audience enjoy the show. Then, to answer the
question
c world. This show presented scienti
90 % of the people who gave a feedback cited
the innovative or evenmagical aspects of the show. The audience found, however, that
the generated music expressed too frankly the emotion being portrayed (75 % of the
audience) and that the music should have taken more distance with the expressed
emotions and leave more interpretation from the spectator. This criticism is interesting
as it validates the parameters we chose for music generation, but pinpoint the fact that
more artistic choices would have been more relevant. Finally, when asked about the
potential improvements that could bemade, 98%of the people stated that they wished
to witness the evolution of the show, should it be in the scienti
“
What did you like?
”
c content, the cho-
reography, the sound generation, and/or the graphical choices. The audience hence
had a very strong interest in following this collaboration between art and science; this
interest went further, as it sparkled reaction frommedias in the form of articles and an
interview on local television.
11.3.4 Valence and Musical Complexity
According to Livingstone et al. (
2010
), the musical parameters related to the
valence of emotions evoked by music are tonality and complexity. We propose here
harmonic, rhythmic and melodic generative models, as well as a description of their
parameters from the point of view of musical complexity.
The way these models (implemented as modules in MuZICO) communicate
between each other is explained by Fig.
11.4
.
11.3.4.1 Pitch Scales
We suggest a model of generative pitch scales base on the following sequence:
V
nmod
ð
m
þ
1
Þ
h
x
p
U
n
þ
1
¼ U
n
x
U
0
ð
11
:
1
Þ
p
2
N
þ
such that U
0
\
U
n
þ
1
\
x
U
0
where U
0
¼ ff
f
;
and
x
h
where
the number by which the octave is divided to get the
temperament,
f
f
the root frequency of the generated scale,
V
a
m
-dimensional vector,
and V
i
;
i
2
½1
::
m
the
i
ith component of
V
.
We consider that the complexity of a pitch scale depends on the number of
iterations needed to generate it, and on the values of
x
,
h
, and
V
.
For example, let us consider the case where
is the octave ratio,
x
¼ 2
,
h
¼ 12
(tonal occidental
music).
Let be
V
= (7) (generation of the pitch scale by iterating through steps of
fifths),
and
N
the number of iterations. This gives:
