Information Technology Reference
In-Depth Information
possible to play a 3D driving game in a way that is not entirely unlike using other
kinds of controllers, depending on the user
is skill level. Although it is of course
accepted that it is not an interface that in any way competes with a conventional game
controller, it is an approach that with machine-learning based single trial detection
might begin to be usable in such scenarios in the near future. We present results of
trials carried out using user-derived machine learning classi
'
ers later in this chapter.
3.6
P300 DJ
Another simple modi
cation of the speller-based approach is the P300 DJ. In many
ways, this system is the most usable of the ones described here as it assumes a
speci
c interaction paradigm that requires little input from the user, and wherein
musical data
c restrictions.
In the use case presented here, the P300 DJ can use any audio
—
the sounds used by the system
—
have speci
file providing it is
trimmed to a loop. The loop can be of any length or numbers of beats as long as
they are multiples of 4. The system has the capacity to estimate the number of beats
based on basic onset detection compared to the sample length, so long as it can
assume that the audio
file starts and stops on a bar division. It can then ensure that
loops of different tempi can be synchronised.
The BCMI system is straightforward. The user has the opportunity to queue
tracks into a playlist using the P300 Composer interface, but instead of selecting
individual notes, the user is presented with a
finite list of available tracks. The user
can move back and forth, selecting or removing tracks from the playlist, and when
each track is about to complete, the next playlist selection is beat-matched and
automatically mixed in.
A different iteration of this system can be built using the P300 scale player
interface, allowing the user to move within a long list of songs, then deliberately
selecting a speci
c track.
This system was created to work with my commercial audiovisual mash-up tool,
Mabuse, which has a number of features that can be used for creating beat-aligned
interactive music and visuals. Again, as with the design of any P300 BCMI, the
crucial approach is to
find an interaction that is possible with the P300 averaging
technique whilst remaining suitable for the user
s needs. I have performed on a few
occasions with this system, and it can produce as satisfying a set as with con-
ventional methods, as the user need only specify something new very rarely when
compared to other forms of music and sound performance.
'
3.7
P300 Influenced Algorithmic Improviser
Another approach that can provide fertile ground for experimentation is to combine
the P300 selection system with an algorithmic improvisation tool, such as a Markov
model-based melody and chord generator.
In systems such as this, speci
c
