Information Technology Reference
In-Depth Information
Fig. 1.1
With BCMI-Piano
one can play music generated
on the
fly on a MIDI-enabled
acoustic piano with brain
signals. No hands needed
musical devices and software to communicate with each other. MIDI itself does not
make sound. Rather, it encodes commands, such as
. In our
case, MIDI commands controlled a mechanism built inside the piano that moved
the hammers to strike the strings. This resulted in a system whereby a real piano is
played with brain signals. The quality of the sound improved considerably. But still,
we felt that the result was not convincingly musical: the output sounded almost as if
the notes were generated randomly. If we were to demonstrate that it is possible to
play music with brain signals, then system ought to do more than merely associate
brainwave activity with notes. Ideally, the music should result from some form of
musical thinking detectable in the EEG. But the task of decoding the EEG of a
person thinking of a melody, or something along these lines, is just impossible with
today
'
note on
'
and
'
note off
'
s technology.
Thus, I came up with the idea of endowing the machine with some form of
musical intelligence, which could be steered by the EEG. The big idea was to
programme the system with the ability to compose music on the
'
fly, obeying simple
abstract generic commands, which might be conveyed by something detectable in
the EEG. This would not necessarily correlate to any musical thought at all, but it
would at least be a realistic point of departure. For instance, I was aware that it is
relatively straightforward to detect a pattern in the EEG, called alpha rhythm, which
is present in the EEG of a person with eyes closed and in a state of relaxation.
Thus, my team and I moved on to implement BCMI-Piano, a system that looks for
information in the EEG signal and match the
findings with assigned generative
musical processes corresponding to distinct musical styles. We implemented an
arti
cial intelligence system that is able to generate pieces of piano music in the style
of classic composers, such as Schumann, Satie, Beethoven, Mozart and so on. For
instance, if the system detects prominent alpha rhythms in the EEG, then it would
activate assigned processes that generate music in the style of Robert Schumann
s
piano works. Conversely, if it detected an EEG pattern other than alpha rhythms,
then it would generate music in the style of Ludwig van Beethoven
'
'
s sonatas
for piano.
