Identifying Saxophonists from Their Playing Styles - Intelligent Music Information Systems: Tools and Methodologies - page 105

sions (Apel, 1972). In order to understand what
is expressed in a performance it is necessary to
understand how it is expressed.
Understanding and formalizing expressive
music performance is an extremely challeng-
ing problem, which in the past has been studied
from different perspectives (e.g., Bresin, 2002;
Gabrielsson, 1999; Seashore, 1936). The main
approaches to empirically studying expressive
performance have been based on statistical
analysis (e.g., Repp, 1992), mathematical modeling
(e.g., Todd, 1992), and analysis-by-synthesis (e.g.,
Friberg, Bresin, Fryden, & Sunberg, 1998). In all
these approaches, it is a person who is responsible
for devising a theory or mathematical model which
captures different aspects of musical expressive
performance. The theory or model is later tested
on real performance data in order to determine
its accuracy. The majority of the research on
expressive music performance has focused on
the performance of musical material for which
notation (i.e., a score) is available, thus provid-
ing unambiguous performance goals. Expressive
performance studies have also been very much
focused on (classical) piano performance in which
pitch and timing measurements are simplified.
This chapter describes a machine learning
approach to investigate how skilled musicians
(saxophone Jazz players in particular) express
and communicate their view of the musical and
emotional content of musical pieces and how to
use this information in order to automatically
distinguish among interpreters. We study devia-
tions of parameters such as pitch, timing, ampli-
tude and timbre both at an internote level and at
an intranote level. This is, we analyze the pitch,
timing (onset and duration), amplitude (energy
mean) and timbre of individual notes, as well as
the timing and amplitude of individual intranote
events. We consider both performances of musi-
cal material for which notation is available, and
performances for which no notation is available,
that is improvising and playing by ear. We focus
on saxophone performance where timing, pitch

and timbre measurements present a greater chal-
lenge compared to the measurements in piano
performances.
Previous research addressing expressive music
performance using machine learning techniques
has included a number of approaches. Lopez de
Mantaras and Arcos (2002) report on SaxEx, a
performance system capable of generating ex-
pressive solo saxophone performances in Jazz.
Their system is based on case-based reasoning;
a type of analogical reasoning where problems
are solved by reusing the solutions of similar,
previously solved problems. In order to generate
expressive solo performances, the case-based
reasoning system retrieves from a memory con-
taining expressive interpretations, those notes
that are similar to the input inexpressive notes.
The case memory contains information about
metrical strength, note duration, and so on, and
uses this information to retrieve the appropriate
notes. One limitation of their system is that it is
incapable of explaining the predictions it makes
and it is unable to handle melody alterations, for
example ornamentations.
Ramirez et al. (2006) have explored and
compared diverse machine learning methods for
obtaining expressive music performance models
for Jazz saxophone that are capable of both gen-
erating expressive performances and explaining
the expressive transformations they produce. They
propose an expressive performance system based
on inductive logic programming which induces a
set of first order logic rules that capture expres-
sive transformation both at a note level (e.g.,
note duration, energy) and at an intranote level
(e.g., note attack, sustain). Based on the theory
generated by the set of rules, they implemented
a melody synthesis component which generates
expressive monophonic output (MIDI or audio)
from inexpressive melody MIDI descriptions.
With the exception of the work by Lopez de
Mantaras and Arcos (2002) and Ramirez et al.
(2006), most of the research in expressive per-
formance using machine learning techniques