Information Technology Reference
In-Depth Information
11.4
Conclusion and Discussion
MuZICO proved its pertinence, on the one hand by its ability to generate music
based on the physiological measure of vigilance, on the other hand in the generation
of music based on emotions represented in a two-dimensional space, the axes of
which correspond to the notions of arousal and valence.
In the near future, we will thus be able to generate music as a function of
emotions not only expressed by the listener, but also felt by this one, and by this
way optimize in a signi
cant manner the retroaction, improving the music
'
s power
of evocation together with its aesthetic qualities.
Another upcoming stage is to re
ne our sound synthesis engine by putting in
correspondence in a more precise way emotions and lower level musical parameters
(corresponding to the perceptive parameters of generated sounds and used sound
samples), and by a better bounds management, conforming to a URL like name-
space as is the case in the Open Sound Control protocol.
An alternative track for the validation of the project is to use reinforced machine
learning algorithms (Le Groux and Verschure
2010
) in order to determine the most
evocative parameter combinations for particular emotions. Some experiments have
been carried out in this direction using this family of algorithms.
11.5 Questions
1. How does MuZICO process emotions?
2. How is the IAPS system used in the experiments on emotion recognition?
3. What emotional components does MuZICO use as dimensions of its emotional
space?
4. Which musical parameters does the result of arousal analysis control?
5. According to Schmitt, how could valence be seen in EEG brainwaves?
6. According to Livingstone, which musical parameters are related to valence?
7. What is soni
cation?
8. How does the LDA algorithm work?
9. What is the difference between emotions computed from an EEG headset and
emotions computed from a motion-capture suit?
10. In the Eq. (
11.1
), what values of
x
,
h
, V, and how many iterations are needed
to build a diatonic scale?
Acknowledgments This research was carried out in the context of the SCRIME (Studio de
Cr
é
ation et de Recherche en Informatique et Musique Electroacoustique, scrime.labri.fr) project
which is funded by the DGCA of the French Culture Ministry, the Aquitaine Regional Council.
SCRIME project is the result of a cooperation convention between the Conservatoire of Bordeaux,
ENSEIRB-Matmeca (school of electronic and computer scientist engineers) and the University of
Sciences of Bordeaux. It is composed of electroacoustic music composers and scientific
researchers. It is managed by the LaBRI (laboratory of research in computer science of the
University of Bordeaux,
www.labri.fr
)
. Its main missions are research and creation, diffusion and
