Computer Music Journal: Music Information Processing
MIT Press, Cambridge MA, 2002
Volume 26, Nr. 2, Summer 2002.
126 pp. illus. $14.00.
ISSN 0148-9267
Reviewed by Stefaan Van Ryssen
Jan Delvinlaan 115, 9000 Gent
stefaan.vanryssen@pandora.be
This issue of Computer Music Journal contains four full articles related
to
problems of musical data processing and one about a tuning application
using
MIDI. There are interesting sections with news from the computer music
world, reviews of cd's, publications and events and new product information
as well.
David Huron of Ohio State University gives an interesting overview of
his
Humdrum Toolkit, a suite of command-line software tools for processing
musical data in predefined or custom representations. The most striking
features of Humdrum are its versatility and the fact that it easily
communicates with other music software. It also allows scientists and
musicians to use their own data representations, which makes it attractive
as an addition to the tools they are already acquainted with. For beginning
computer users, its command line interface may well be a bit complicated.
Brian Pardo and William P. Birmingham from the University of Michigan
describe some algorithms for chordal analysis. Any music student will
know
how difficult it can be to identify chords in classic scores. The task
can
be described as a two-step process: identifying harmonic 'segments'
and
labelling the segments with the right chord. The authors have defined
metrics to evaluate the performance of their algorithms, thus laying
the
foundations for further work on improved algorithms. The article certainly
will be of interest for those who are using or developing software for
teaching harmony as well.
Dominic Mazzoni and Roger B. Dannenberg from Carnegie Mellon University
describe 'A Fast Data Structure for Disk-Based Audio Editing'. In this
highly technical, engineering-oriented article, they describe an approach
which combines the advantages of two basic types of music editors: the
non-destructive sample-mixing kind (e.g. Cubase) and the waveform
reprocessing kind (e.g. CoolEdit).
Their program, nicely called 'Audacity', involves the storage of blocks
in
nodes and keeping track of sequences of these blocks in graphs and trees.
Editing operations are usually performed on at most a few blocks or
a few
nodes at a time. A history of changes can be kept in a memory stack,
allowing a theoretically unlimited number of 'undoes' without wasting
too
much storage space.
Audacity is freely available for download at audacity.sourceforge.net
.
Eduardo Reck Miranda from the SONY Computer Science Lab in Paris explores
'Emergent Sound Repertoires in Virtual Societies'. He simulated the
emergence of music in a freely communicating society. This is a most
interesting contribution for those who are interested in robotics, the
study
of emergent behaviour and the origin of communication systems. Miranda
clearly explains the motivation for his research, his methodology and
the
implementation of the virtual community.
In short: agents in the community are fitted with sound production and
perception modules. They start with an empty memory and interact in
a game
of 'uttering' sounds, imitating them and reinforcing successful sounds
and
forgetting unsuccessful ones. By playing several hundreds of rounds
of games
in a row, a community of twenty agents gradually evolves a shared repertoire
of about five to six 'meaningful' sounds. Using these and a simple grammar,
it is possible to let them create a proto-conversation.
Whether the resulting sound can be called music is doubtful, but that
is not
really the issue. 'Music' is a name we give to a subset of all our
communicative interactions. The amazing thing is that this and similar
experiments clearly show the emergence of a limited repertoire of sounds
as
a result of and a prerequisite for social bonding in a society of
individuals whose main traits are an auditory apparatus and a strong
drive
to imitate and communicate.
"Groven.Max: An Adaptive Tuning System for MIDI Pianos" by
David Leberg Code is the odd one out in this issue. Code, currently
at Western Michigan University, has developed some of the ideas and
experiments of Eivind Groven, a Norwegian composer and ethnomusicologist
who strove to build instruments in just intonation. Groven.Max is a
MIDI patch designed to distribute the notes played on one keyboard to
three differently tuned pianos in order to emulate a tuning system using
pure intervals from the harmonic series. In this article, historic and
theoretical backgrounds are summarized and a general outline of the
system is given. The Groven Piano premiered in Oslo in April 2001. Anyone
interested in intonations and organology will find this an inspiring
read.