ArticlePDF Available

Abstract and Figures

Electroacoustic music analysis is a complex and heterogeneous discipline depending on one musical genre which includes a large typology of subgenres: from tape music to computer music, from concrete music, to mixed music, live electronic music, laptop music, etc. Even though there are personal approaches, which causes musical analysis to be a delicate and subjective discipline, some main trends can be outlined: some analysts skip the technological dimension and base their work on perceptual dimension; other ones deepen a genetic approach. Computer science applied to sound features' extraction begins being interested to this music with promising perspectives. Any approach is worth being considered in order to create an interdisciplinary research area in electroacoustic music analysis. In this paper, the point of view is the musicological one. The goal is to outline a general survey of different musicological and computational approaches. Each of them is partial. What musicologists and scientists now need is to cooperate and share different competences. Interdisciplinary character of future studies is fundamental.
Content may be subject to copyright.
This is the first draft for the plenary talk to be presented at SMC 05 in Salerno, Italy, on November 25. Please do not copy or cite this version and refer
to the Proceedings published on paper.
Laura Zattra
University of Padua
Department of Visual Arts and
Electroacoustic music analysis is a complex and
heterogeneous discipline depending on one musical
genre which includes a large typology of subgenres:
from tape music to computer music, from concrete
music, to mixed music, live electronic music, laptop
music, etc.
Even though there are personal approaches, which
causes musical analysis to be a delicate and subjective
discipline, some main trends can be outlined: some
analysts skip the technological dimension and base
their work on perceptual dimension; other ones deepen
a genetic approach. Computer science applied to sound
features’ extraction begins being interested to this
music with promising perspectives. Any approach is
worth being considered in order to create an
interdisciplinary research area in electroacoustic music
In this paper, the point of view is the musicological
one. The goal is to outline a general survey of different
musicological and computational approaches. Each of
them is partial. What musicologists and scientists now
need is to cooperate and share different competences.
Interdisciplinary character of future studies is
According to the musicological encyclopaedia New
Grove Dictionary of Music and Musicians, musical
analysis is the “resolution of musical structure into
relatively simpler constituent elements, and the
investigation of the functions of those elements within
that structure” [1]. A more recent New Grove’s
definition states that “analysis is the means of
answering directly the question ‘How does it work?’. Its
central activity is comparison. By comparison it
determines the structural elements and discovers the
functions of those elements” [2]. The aim of the
analysis is therefore empirical and based on the musical
phenomenon itself, instead of external aspects
(biographical, political, social, educational, etc.).
Since the material nature of music is difficult to define
(as it is not a tangible entity), the object of the analysis
must be clearly defined; it can be the score or the sound
image, the mental image of the composer or its
performance, etc. The definition of the limits and
parameters of any analytical operation is therefore
essential in order to make a useful work for the
comprehension of music and in order to possibly avoid
– or declare – the subjectivity of the analytical process.
“The very existence of an observer – the analyst – pre-
empts the possibility of total objectivity. No single
method or approach reveals the truth about music above
all others” [2, 528]. Any analytical method must start
with the declaration of the dimension to analyse: the
microstructure or the macrostructure, the medium level,
the timbre, etc. [3].
Electroacoustic music analysis is still problematic
because these dimensions are blurred and there is no
stable compositional theory which could reflect and
guide the process of listening (fulfilment or frustration
of expectations [4]).
Nevertheless, a series of different analytical approaches
applied to electroacoustic music starts to reach
interesting results. Among musicological works, we
find two main attitudes: one emphasizes a purely
perceptive method based on Listening as a mean to
comprehend a musical work, and justifies its view by
saying that electroacoustic music is made “for listening
and by listening” [5, 158]. The other approach aims at
comprehending the genesis of the musical work, and
investigates compositional sketches, scores, different
type of technological data, etc.
Computational analysis starts to take interest in
electroacoustic music and could apply its present results
in automatic extracting features of sounds,
classification and segmentation, for musicological
analysis, its representation and transcription. Because
the task of perception is complex and subjective,
computer can be of great interest permitting to reach a
greater objectivity through modelling.
This work can be viewed as one attempt to organize the
knowledge and competence electroacoustic music
community has reached in analysing its product.
Electroacoustic music analytical approaches (a) can
consider the sound material of one piece just as pure
sound, and refer to it as sound objects which recall
something existent in nature, or which have a physical
identity different from other sound objects, or (b) they
can base their work on the knowledge of the
technological environment used to generate the sound.
In both cases, we think that the knowledge of the
historical period and instruments typical of the musical
repertory is fundamental, because electronic music
equipment, with their potentials and limits, influence
the typology of sound, the compositional process, the
performance and the listening. Even if he is not
interested in the real process of the technical realization
of sound, the analyst should consider these aspects in
order to understand at least the compositional and
perceptive problems.
Each historical period has been influenced by
technology and by musical context. For example, the
input of Concrete Music is the recorded sound. The
principle instruments of musical writing, at the time of
Pierre Schaeffer, were the following: the microphone,
the reverberation (initially a room or a box provided
with loudspeakers and microphone), the tape and the
Phonogène, an instrument with keyboard which
permitted to vary the reading velocity of the tape (the
ancestor of modern samplers). Computer and digital
software now concentrate the whole process in one
unique and flexible system (except for the microphone,
still necessary to tap the sound). These instruments
influence evidently the typology of sound
transformation which can be: quotation, loop, echo-
reverberation, filtering, modulation of one or more
parameters, change of reading velocity of the tape (time
stretching, transposition), tape reversing, spatialization
On the other hand, electronic music developed inln
in Germany at the WDR studio works from the opposite
point of view, with wave and noise generators. Key-
word is creating sound, not recording. Compositional
approach derives from Serialism and electronic works
must be analysed following this musical vision.
Computer music exploits software created for sound
synthesis and its transformations. From MUSIC
software to more recent programs, digital systems are
nowadays very numerous and show different
characteristics which influence sound, compositional
practice and therefore music. Compositional techniques
depend on the composer’s approach and analysis is
related to the precise works.
The first experiments of Mixed Music were made by
E.Varèse (sert, 1954) and in Köln (Mantra by
K.Stockhausen, 1969). These works represent the two
main approaches of mixed music: works for tape and
instruments (the interpolation of electronic and acoustic
timbre is the musical goal) or works whose acoustic
sounds are modified in real time.
Acousmatic music uses recordings of everyday sounds,
recordings of instruments and synthesized sounds. They
are combined in their raw forms or processed, then
mixed together and recorded to a fixed medium (usually
CD or tape).
These are just some definitions of different subgenres of
a large and heterogeneous musical world which is
electroacoustic music. The most evident problem which
arises from this is that a musical work, depending on
the technology used, may be stored in different formats.
The same music piece can be represented, for example,
by a symbolic notation of the score, by a sequence of
time-stamped events corresponding to pitched and un-
pitched sounds, by the recording of an acoustic
performance, by a series of operational or digital data
corresponding to the calculation of the software for
digital synthesis.
Up to now, nobody has really succeeded in considering
at the very same time all this heterogeneous
documentation for the analysis. We can see it from the
following analytical approaches.
3.1. Listening Analysis
Listening musicological analysis is based on the idea
that electroacoustic music does not benefit of a unified
representation code relying sonorous text with the
compositional work of the composer. For this reason,
the representation of the listening experience becomes
the only mean to understand and study this music.
Musicological literature shows individual analyses
(made by musicologists or composers) or, in some rare
cases, works which compare different tests of analysis.
The main, by now historical, studies we can mention
are the major studies by Denis Smalley [6], Simon
Emmerson [7], Michel Imberty [8], François Delalande
[9], Francesco Giomi and Marco Ligabue [10].
3.1.1. Pierre Schaeffer
These studies are all inspired by the pioneer researcher
Pierre Schaeffer [11], where the attention for the inner
structure of sound finds its realization in the concept of
sound object, objet sonore. The book Traité des objets
musicaux written in 1966 is an extraordinary work of
700 pages, in which Schaeffer establishes the Concrete
Music. Here Schaeffer studies the nature of the sound,
which is the timbre and its form (attack, body, decay).
He makes a catalogue of sounds objects (solfège)
through five operations: typology, morphology,
characterology, analysis, synthesis. Typology and
morphology are complementary: morphology indicates
the quality of sound (description: mass, timbre
harmony, dynamics, grain, allure, melodic profile, mass
profile), typology classify it (according to some criteria:
mass / facture, duration / variation, equilibrium /
originality). He finds 29 sound typologies, divided in
well-balanced objects, redundant and eccentric objects.
Each object is described by a table with 7 vertical
morphological descriptions and 9 horizontal typology
descriptions. For this reason, Schaeffer’s analysis is
called Typo-Morphology.
3.1.2. François Delalande
From the ‘70s François Delalande deepens Schaeffers
approach. He usually makes auditory tests for analysing
perception of music and thinks that since listening is
subjective, there is no definitive analysis. Listening
conducts are actions finalized to a specific goal [12].
After testing listeners, Delalande analyzes data
collected and different listening conducts, traces main
listening conducts which correspond to a more objective
analysis. Subjectivity reduces itself considering that
each real conduct is a combination of at most 3
different listening-types: taxonomic type, emphatic type
and figurative type [9].
3.1.3. Denis Smalley and Spectromorphology
Denis Smalley thinks that analysis of electroacoustic
music “has to start with the perceptual choices of the
listener-analyst who selects pertinent criteria. I must
acknowledge that each listener, including the analyst,
will make an individual, maybe unique reading of a
work, but I must also acknowledge that individual
readings are variations springing from shared,
acculturated, human perceptions and needs” [14, 423].
He theorizes the notion of source bounding that is the
extrinsic signification of sound material within a
musical piece. He studies it through spectro-
morphology that is the analysis of the characteristics of
sound “through time – types of sounds and their
shaping”. Spectrum is defined by note, node and noise.
Note can be exact, harmonic or enharmonic [13].
Figure 1. Movement and growth process of sound
(Smalley 1996).
Spectrum realizes itself in a temporal dimension
(Motion) and can have 3 archetypes: attack-impulse,
attack-resolution, continuous sound. From that, one can
studies all typology of movement and growth of
(harmonic or enharmonic, electronic, etc.) sound [6].
3.1.4. Simon Emmerson. Sound-image-syntax
Simon Emmerson too, develops the image sound evokes
by listening experience and introduces the concept of
mimesis of natural or abstract sounds. Electroacoustic
music works can be placed in a more aural discourse
or, on the contrary, in a mimetic discourse, according to
the abstraction of imitation. Musical form, in its turn,
can be based on an abstract syntax or abstracted syntax
(extrapolated syntax) [7].
Figure 2. Simon Emmerson musical discourse’s
Simon Emmerson is important for being the first
composer-musicologist who published, in 1986, the first
book on the problem of the language and analysis of
electroacoustic music [15].
3.1.5. Stéphane Roy’s recent book
In 2003 Stéphane Roy published a book devoted to
Listening Analysis [16]. He focuses on acousmatic
works influenced by Schaefferian theory. The book is
divided in two parts:
1) the first one presents various methods applied to
electroacoustic music (Schaeffer, Jean-Jacques
Nattiez, Christiane Ten Hoopen, Delalande, GRM
works, Henri Chiarucci, Robert Cogan, Wayne
Slawson, Francesco Giomi and Marco Ligabue,
John Dack, Denis Smalley and Andrew Lewis);
2) the second part shows 5 approaches which are not
specifically though for this music but could be
useful. These are the paradigmatic analysis by
Nicolas Ruwet, the generative analysis by Lerdahl
& Jackendoff, the implicative methodology by L.B.
Meyer, and Roy functional own analysis.
Roy’s approach is one based on l’analyse du niveau
neutre (abbreviated ANN – analysis of the neutral level,
considering Nattiez’s definition). All analyses in the
second part are applications of different methods to the
same piece (Points de fuite by Francis Dhomont), and
follow a description of their context, commencing with
the creation of a score identified with the neutral level.
Roy identifies some sound ‘unities’ of the piece. Its
representation is bi-dimensional and corresponds to 2
main typologies of sound (the third is a sub-genre of the
second). Internal morphology of sounds can be periodic,
sound-noise, complex (noise).
Figure 3. Roys typology of sounds for the ANN (Roy
Roy’s own personal approach is interesting because he
shows the so-called Functional Grid with symbols of 45
functions classified in 4 main categories (orientation,
stratification, process, rhetoric). The goal is to
comprehend the musical unities pertinent for the ANN
analysis of a musical work.
Figure 4. Stéphane Roy’s Functional Grid for the
Analysis at the Neutral Level.
As Leigh Landy says, an aspect that is perhaps
unexpected is Roy’s rejection of the sonogram or
Acousmographe as a point of departure at the neutral
level. Roy’s reasoning is that acoustics or
psychoacoustics-based research tends to base itself on
the isolation of parameters. In his view, sound
parameters interfere with one another too much in a
complex synergy to be treated separately as far as
electroacoustic music analysis is concerned.
3.1.6. Spectrograms
New images of musical sound (1984) is an important
book for this discipline, where Robert Cogan and Pozzi
Escot used spectrograms for analysis, as a description of
the realization process rather than perceptual [17].
They were enthusiastic for sonogram capacity to
represent music: “only now, through a new synthesis of
scientific and musical analysis, can we begin or probe
the sonic enigma. Photographs of the spectral formation
of musical works provide a bridge that makes a new
understanding of sound and music, sound in music,
possible”. Analyst is still important, anyway, for
understanding sonogram’s meaning: “spectrum photos
display sonic formations vividly, but they do not quite
speak for themselves (a tempting illusion). The
commentaries, therefore, direct the reader’s attention to
those elements that are essential for an understanding
of the photos”.
3.1.7. Multimedia representations and
electroacoustic music
Other studies aim to graphically represent the electro-
acoustic music flux in multimedia contexts, starting
from the musicologist’s personal listening. All these
works aim to describe the listening in order to
understand the musical structure and/or timbre. They
sometimes use acoustic representation tools (time-
amplitude representations, spectrograms, sonograms).
We just cite some of them, being impossible to follow
all the development of this type of works.
Acousmographe is a program for the annotation of
electronic music designed by the ‘Groupe de Recherche
Musicale’ in Paris created by Pierre Schaeffer in the
late 50's. This program displays the waveform and a
sonogram that can be indexed and annotated by the
user. The main feature is the possibility to manually
add a graphical symbolic representation to represent a
segment selected by the user. Recently, an algorithm
that retrieves segments perceptually similar, to speed up
the annotation, has also been integrated.
« Écoutes signées » is a project developed at Ircam in
Paris since 2003 (
) [19]. It aims
at formalizing listening praxis and develops general
computer tools for helping musical listening; it is
thought in particular for electroacoustic music. « A
‘signed listening’ is a hypermedia product which aims
at making a personal and original way of listening
transmissible […], by suggesting types of graphic and
acoustic representations and manipulations of music based
on a preexisting listening practice ». The ultimate goal of
the project is to develop « a generalized tool of computer-
aided auditory exploration that will facilitate and
deepen an organized listening of sound in all its
diversity ». Until now, the team has presented some
analysis based on mixing down some very short loops
of electronic commercial music and some more general
analyses which shows personal annotations of the
listener (a composer).
INA-GRM is still developing the project Portraits
Polychromes, with on-line materials and edited books
devoted to avant-garde composers, their musical role
and works ( Between them there are electroacoustic
composers: François Bayle, Ivo Malec, Bernard
Parmegiani, Gilles Racot, Jean-Claude Risset, Luc
Ferrari John Chowning). Recently GRM published the
interactive CD-ROM La musique électroacoustique
[20], presenting the analysis of 6 works by composers
themselves and 11 examples of qualified listeners. The
CD-Rom shows different approaches from the aesthesic
approach (listening) to the poietic approach (genetic).
Figure 5. Incipit of Bernard Parmegiani’s De natura
sonorum [20].
Figure 6. Bernard Parmegiani’s De natura sonorum
The MIM (Laboratoire Musique et Informatique de
Marseille) (
) works on the
development of Temporal Semiotics Units with which it
studies musical works.
3.2. Genetic analysis
This type of analysis studies the compositional process
[21], or uses computer data as objective material to be
analysed; one of the first studies was Lorrain’s analysis
of Inharmonique by Jean-Claude Risset [22]. These
researches are still pioneering.
This approach was one of the goals of my PhD
dissertation, which analyzed 6 works produced at Ircam
in Paris and at CSC – Centro di Sonologia
Computazionale in Padova [23] [24]. What seemed
fundamental to me for the study of this music, was the
recuperation of the different material used during the
creation of a work: most importantly, computer scores
(MUSIC software data), but also rough sketches,
different recordings, published articles, etc. Reading the
computer data allows an understanding, on the one
hand, of the structure and the sonority of the
instruments, and on the other, their temporal
development within the piece. An evaluation of data
lists permits the knowledge of the ways the computer
was used in the compositional process. Its value can
only be defined whilst taking into consideration the
different tasks performed by the P-fields in MUSIC
software: initialisation, local variables, timbre
transformation, etc.
This approach is fundamental, in my vision, for
analysing electroacoustic music in general, and above
all mixed works. Here interpolation of electronic and
acoustic sounds is the musical principle and the
hybridization affect perception. The analysis of the
support (tape or CD, etc.) is not sufficient to
discriminate sounds, and needs the deep knowledge of
the electronic technology.
Computational analysis exploits the ability of a
computer to analyze sounds and recognize patterns
within a musical piece in a perceptually and musically
meaningful manner. Up to now, significant results have
been obtained for Western tonal music and for popular
This discipline dates back to the first researches, in the
‘70s, dedicated to the analysis of sound. We can cite
CHANT software developed at IRCAM in Paris,
Moorer’s work on segmentation and analysis of sounds
[35], or McAdams and Bregman researches [36].
During the ‘80s, important outcomes derived from
physical models study, whose purpose was to obtain
representations of sounds in order to completely re-
synthesize them or, at least, to obtain valid synthesis
from a perceptual point of view. Sound analysis aimed
at not losing any acoustic characteristic of timbre.
From the ’90s, research has devoted its efforts in Music
Information Retrieval studies, whose goal is to extract
characteristics of sound and, to do it, to discard
irrelevant aspects of it (different steps of analysis
consist in pre-elaborating the sound with noise
reduction, equalization, etc.; selecting frames,
extracting characteristics, post-elaborating).
Computational auditory scene analysis is also
challenging, because modeling the human auditory
system has proved to be very difficult, and promise to
be a long-term interdisciplinary research.
The main problem computational analysis encounters,
is the fact that electronic sounds have heterogeneous
properties which are very different to regular
instrumental harmonic sounds. Electronic sounds are
not coded (they depend from the system), they have not
a natural perceptive characteristic (except for sampled
sounds) their behaviour is not regular and sometimes
they rather seem (according to analytical community) to
be objects to be consider in their unity rather then in
their internal spectral change.
4.1. Music Information Retrieval and Electroacoustic
Music Information Retrieval (MIR) is an
interdisciplinary research area which has grown out of
the need to manage various digital collections of music
and to develop rational methods for managing,
preserving, accessing, researching, etc. this type of
musical material [25]. The general difficulty to manage
music is that it is not a simple object made by simple
and low characteristic but, on the contrary, it show
difficult-to-extract layers of significance, such as
harmony, polyphony, and timbre. MIR works for
developing techniques for extracting these high-level
features. This discipline is interesting because it is a
“content-based” research, to distinguish it from digital
and pre-digital approaches founded on manually-
produced metadata of bibliographic and related
varieties. It aims at automatically extracting features
concerning fine-level transcription of events and broad-
level classification of pieces. The difficulty this
approach encounters is the fact that, for working, it
must previously know the content it is analyzing.
Stephen Downie identifies seven facets of music and
notes having a significant effect on how systems can
retrieve music. These are pitch, temporal, harmonic,
timbral, editorial, textual, and bibliographic [26]. A
large amount of literature exists on the extraction of
rhythmic descriptors from musical data, from symbolic
data, audio or compressed audio. Other dimensions,
above all timbre, are far to be resolved.
Unfortunately, the music used in MIR studies is
predominantly traditional Western music and popular
music and that was the object of the overwhelming
majority of papers presented in the last ISMIR
conferences. Nevertheless, something is moving
concerning electroacoustic music analysis.
4.1.1. An hypothetical method for acousmatic music
David Hirst presents at ISMIR some remarks applied to
acousmatic music, derived from the synthesis of top-
down (knowledge driven) and bottom-up (data-driven)
views [34]. He develops his research starting from
Gygi’s [37] and Howard and Ballas’ [38] experiments
on some sonic instances of sounds played to listeners
and to be classified, and on Bregman studies on timbre
Figure 7. Hirst’s network of relations in the
interpretation of acousmatic music (Hirst 2004).
Nevertheless the paper presents some theoretical
consideration on a hypothetical method. Hirst tells that
what remains to be completed, is to test the method on
representative repertoire works. So, we can not judge
the validity of this approach.
4.2. Description, segmentation, classification
4.2.1. Morphological description of sounds. Julien
Thinking over the deficiency of the MIR studies based
on electroacoustic music, Julien Ricard has worked on
electronic sounds, noises and sounds that have no
identifiable origin, which are used a lot in
contemporary music and sound post production for
video or cinema [40]. He analyzes, describes and
automatically classifies a corpus of separated electronic
sounds, that is not within a musical fluxes. This work
(developed at the Universitat Pompeu Fabra in
Barcelone, Spain) was initiated in the context of the
cuidado-e.html) on audio content description and has
been carried on for the AudioClas project
) on automatic classification of
sound effects. He starts from the concept of typo-
morphology by Pierre Schaeffer [11]. Ricard establishes
a descriptive scheme, amplifying Schaeffer’s
classification, for the evaluation of the electronic sound
objects. He gives each sound a numerical value or a
category: subjective duration, loudness, pitch,
pitchness, roughness, dynamic profile, attack,
brightness, spectral fluctuation, and other (metallicness,
richness, etc.). From this classification he extracts the
pertinent characteristics of each sound and traces a
complex scheme for the morphological description a
sound objects. Some low-level descriptors automatically
digitalize each characteristic of the sound.
Figure 8. Block diagram of the automatic
morphological description system (Ricard 2004).
This morphological descriptor has been tested on 200
musical examples. The sounds were manually labelled
according to the classes defined earlier and then tested
on each descriptive criterion. Classification
performance is generally good, but weak with complex
sounds (with simultaneous components or noise,
performance of 50%). These results are still
problematic. Nevertheless they are very important and
promising for the study of electroacoustic music.
4.2.2. Segmentation and classification of audio files
Another experimental work has been carried out by
some students in Padova University during an internal
research on segmentation and classification of audio
files [41]. They start from the works of Wyse and
Smoliar [42] (on the categories of music, speech and
‘other’ based on frequency analysis), Kimber and
Wilcox [43] (on speech, silence and non-speech,
thought for segmenting conference’s talks), and Pfeiffer
[44] (analysis of amplitude, frequency and fundamental
frequency of an audio file for segmenting auditory
scene, including noise: shots, weeping, etc.). They
analyzed an audio file prepared on purpose for the
experiment, which contained a series of vowels, silence,
metronome click, a fragment of song, instrumental
sounds, and other letters. Using temporal windows,
they calculated features (by analysing short time
average energy, average zero crossing rate, short time
fundamental frequency), researched rough variations,
unified and damped all data. They manually classified
and post-processed the data collected
through a
morphological and statistical analysis.
This research still
encounters serious problems with enharmonic sounds.
Moreover, it works only with a series of separates
sounds because when two sounds overlap in time and
frequency, it is extremely difficult to resolve them.
4.2.3. Recent applications
Up to now, it is clear that most of the attention in this
area of studies is paid to limited type of sounds, and
more particularly to the design of automatic music
transcription systems of traditional western tonal music.
Electronic and computer sounds, noises or sounds that
have no identifiable origin, can hardly be handled.
Some recent dissertations realized at Padova University
(and presented during this SMC conference) addressed
the analysis of computer music signals within a real
piece of music. They provided some first attempts to
adapt existing analytical frameworks to electro-acoustic
music, in order to overcome the lack of specific
objective methodologies for no traditional western
music [48]. Their results are problematic but promising
though. The helped in understanding the fact that audio
engineering must collaborate with musicology for
accelerating the comprehension of problems
musicological and audio analysis has.
Visualisation of electroacoustic music
Good general-purpose transcribers in traditional
notation do not even exist, let alone for electroacoustic
music, a musical genre whose notation is very difficult.
If we consider problems computational transcribers of
Western music encounter, we can cite accuracy and
flexibility which are not comparable to human
musicians; extraction of discrete notes is still difficult;
drums, percussive sounds are impossible. Some
promising results are evident but just for limited-
complexity music. But music is a complex and
polyphonic sonorous phenomenon. So, if it seems that
musicological (higher-level) models are necessary to
further improve the transcription accuracy, on the other
hand transcription is necessary for musicological
analysis. Therefore, this discipline has to be improved
for and from both approaches.
Anssi Klapuri, underlines this difficulty by saying that
“an important fact about music transcription is that it is
difficult. […] The problem is really not in finding fast
computers but in discovering the mechanisms and
principles that humans use when listening to music.
[…] Anyone who claims to have a quick solution to the
polyphonic transcription problem, or a single
mechanism that solves the problem once and for all is
mistaken. The human brain combines a large number of
processing principles and heuristics. We will be
searching for them for years, perhaps even decades,
before arriving at, say, 95% of a skilled musician’s
accuracy and flexibility [46].
Anderson Mills tries to apply psychoacoustic audio
analysis techniques to electroacoustic music for the
purpose of visualization [32]. His goal is to create
algorithms which use models of human hearing to
extract audio properties from recorded electroacoustic
music (based on human hearing model algorithms: Dr.
Patterson’s Auditory Image Model, Drs. Meddis and
Hewitt’s Correlogram, Dr. Ellis’ Weft works, and then
visualise all data in a graphical score [47].
Figure 9. Anderson Mills’s graphical score prototype
Mills work just shows some experiments which could
be very interesting for the future research.
Next year, musicological community will celebrate the
anniversary of the book The language of
electroacoustic music, London, Macmillan, 1986 edited
by Simon Emmerson, the first book where the problem
of the analysis of electroacoustic music was clearly
defined. In these years, musicologists and sound
engineers have been more and more interested in this
problem. However, ggiven the rapidly increasing
number of electroacoustic music scholars, as well as the
fact that large web-based music collections are
continuing to grow in size exponentially (experimental
and commercial electronic music), it is obvious that this
amount of research must be organized, collected and
organized in a more systematic way. Electroacoustic
music community needs now to start organizing these
different approaches, in order to exploiting their results
for the creation of common instruments for the analysis.
This is the first, very urgent, need.
It is also time to reflect on all different approaches,
evaluating their merits and defects and maybe to mix
them. The Listening Analysis depends on the
competence of each listener; the common idea is that
transcription cannot have the ambition to explain the
musical object as a whole, but just to underline some
aspects depending on the personal approach of the
listener. However, its desire to categorize sound objects
independently from their technological realization
could be the right way to join a common notation of this
music. This group of musicologists could help MIR
researchers who have already pointed out, that MIR
research programs should also agree upon evaluation
measures [25]. Retrieval accuracy and system
effectiveness should be measured using clearly
delineated, agreed-upon methodologies and reported
consistently across studies.
Electroacoustic music community devoted to analysis
should reflect, before starting any analytical operation,
on the following fundamental questions and concepts,
which concern me closely:
1) analysis is set up on the goal it is looking for. This
can be:
the preservation of a musical heritage for
permitting the re-synthesis or preserving the
performance praxis, etc. (for practical reasons, e.g.
obsolescence of systems, or for didactics);
tracing graphical scores for helping the listener in
the comprehension of this music;
tracing automatic scores for helping the
musicologist in the investigation of structural
dimensions of the musical piece;
automatic classification of electroacoustic music for
the web search and/or for determining
electroacoustic music genres, now still flowing;
2) Which are the main deficiencies and capabilities of
actual research?
All actual analytical methods are partial. What I think
musicologist and scientists need is to cooperate and
share different competences. The musicologist is
interested in comprehending the sonorous identity
which is the electroacoustic music work. Its knowledge
of the fragmented network of agents and processes
concerned in making a piece, together with the
competence of computer scientist, could bring the
electroacoustic music analysis to a more clearly defined
Interdisciplinary character of future studies is
fundamental. The future procedures for the analysis of
electroacoustic music should be derived from the
synthesis of top-down (knowledge driven) and bottom-
up (data-driven) views derived from different
[1] Bent, I. "Analysis," The New Grove Dictionary
of Music and Musicians, ed. Stanley Sadie,
London: Macmillan, 1980, vol. 1, pp. 340-88.
[2] Bent, I. - Pople A. "Analysis", The New Grove
Dictionary of music and musicians, vol. 1, II
edition, Macmillan Publishers, 2001, pp. 326-
[3] Dalmonte, R. "Teoria e analisi", Enciclopedia
della musica. Il Novecento, Torino, Einaudi,
2001, pp. 659-676.
[4] Meyer, L. B. Emotion and meaning in music,
University of Chicago Press, 1956.
[5] Delalande, F. "Pertinence et analyse
perceptive", La revue musicale, pp. 158-173,
1986, and Cahiers Recherche/Musique n. 2,
Paris, INA/GRM.
[6] Smalley D. "La spettromorfologia: una
spiegazione delle forme del suono (I)", "(II)",
M/R Musica/Realtà n. 50 1996/3, n. 51
1996/3, pp. 121-137 / pp. 87-110, LIM -
Quaderni di Musica/Realtà, 1996.
[7] Emmerson S. "The relation of language to
materials", The language of electroacoustic
music, London, MacMillan Press, pp. 17-39,
[8] Imberty M. "Continuità e discontinuità",
Enciclopedia della musica. Il Novecento,
Torino, Einaudi, pp. 526-547, 2001.
[9] Delalande F. "Music analysis and reception
behaviours: Sommeil de Pierre Henry", Journal
of new music research, vol. 27, no. 1-2, pp.
13-66, 1998.
[10] Giomi, F. & Ligabue, M. "Un approccio
estesico-cognitivo alla descrizione dell’objet
sonore", R. Dalmonte - M. Baroni (eds.),
Secondo convegno europeo di analisi
musicale, Trento - Università degli studi di
Trento, pp. 435-448, 1991.
[11] Schaeffer P. Traité des objets musicaux. Essais
interdisciplinaires, Paris, Seuil, 1966.
[12] Delalande, F. "L’articulation interne/externe et
la détermination des pertinences en analyse",
Observation, analyse, modèle: peut-on parler
d’art avec les outils de la science? (Chouvel J-
M-Levy F.), Paris, L’Harmattan – IRCAM
Centre Georges Pompidou, pp. 175-194, 2002.
[13] Smalley, D. "Spectro-morphology and
structuring processes", The language of
electroacoustic music (S. Emmerson ed.),
London, MacMillan Press, pp. 61-93, 1986.
[14] Smalley, D. "Can electroacoustic music be
analysed? ", R. Dalmonte - M. Baroni (a cura
di) - Secondo convegno europeo di analisi
musicale, Trento - Università degli studi di
Trento, pp. 423-434, 1991.
[15] Emmerson, S. The language of electroacoustic
music, London, MacMillan Press, 1986.
[16] Roy, S. L’analyse des musiques
électroacoustiques: Modèles et propositions,
Paris, L'Harmattan, Univers Musical, 2003.
[17] Cogan, R. – Pozzi, E. New images of musical
sound, Cambridge, Massachusets and London,
Harvard University Press, 1984.
[18] Delalande, F. Le condotte musicali, Bologna,
Clueb, 1993.
[19] Donin, N. "Towards organised listening :
some aspects of the ‘Signed Listening’ project,
Ircam", Organised Sound, 9(1), 2004, pp. 99-
[20] Bayle, F. La langue inconnue. INA/GRM, La
musique électroacoustique, édition, coll. Musiques Tangibles, Paris,
[21] Analyses musicales (1996-2005),
[22] Lorrain, D. "Analyse de la bande magnetique
de l’œuvre de Jean-Claude Risset
Inharmonique" – Rapports IRCAM 26/80,
[23] Zattra, L. Science et technologie comme
sources d’inspiration au CSC de Padoue et à
l’IRCAM de Paris, Ph.D. Thesis, Paris IV-
Sorbonne – Trento University, 2003.
[24] Zattra, L. "Searching for lost data : outlines of
esthesis-poietic analysis", Organised sound,
vol. 9, no. 1, april 2004, Cambridge University
Press, pp.35-46, 2004.
[25] Futrelle, J., & Downie, J. S. "Interdisciplinary
Communities and Research Issues in Music
Information Retrieval", Third International
Conference on Music Information Retrieval
(pp. 215-221), IRCAM, Paris, 2002.
[26] Downie, J. S. "Music information retrieval",
Annual Review of Information Science and
Technology, 37, 295-329, 2003.
[27] De Poli, G. "Analisi dei suoni", dispensa del
corso di Informatica Musicale, Corso di Laurea
specialistica in Ingegneria Informatica (nuovo
ordinamento), Informatica Musicale,
Università di Padova, Dipartimento di
Ingegneria dell'Informazione, 2005,
[28] Bonardi, A. "IR for Contemporary Music:
What the Musicologist Needs". International
Symposium on Music Information Retrieval,
[29] Cope, D. "Computer Analysis of Musical
Allusions". Int. Symp. on Music Information
Retrieval, Bloomington, IN, USA: 83-84, 2001
[30] Kornstädt, A. "The JRing System for
Computer-Assisted Musicological Analysis",
International Symposium on Music
Information Retrieval, Bloomington, IN, USA:
93-98, 2001.
[31] Smiraglia, R. P. "Musical Works as
Information Retrieval Entities:
Epistemological Perspectives", International
Symposium on Music Information Retrieval,
Bloomington, IN, USA: 85-91, 2001
[32] Anderson M., "The Application of
Psychoacoustic Audio Analysis Techniques to
Electroacoustic Music for the Purpose of
Visualization", (PhD work). Powerpoint
presentation no more available on-line: 147th
meeting of the Acoustical Society of America,
New York, NY, 2004–05–25, 2004.
[33] vy, F., Complexité grammatologique et
complexité aperceptive en musique, Étude
esthétique et scientifique du calage entre la
pensée de l’écriture et la perception cognitive
des processus musicaux sous l’angle des
théories de l’information et de la complexité,
Thèse pour obtenir le grade de docteur de
l’EHESS en musicologie (sous la direction de
Jean-Marc Chouvel, Co-direction : Marc
Chemillier), Ecole des Hautes Etudes en
Sciences Sociales Formation Doctorale «
Musique et Musicologie du XXe siècle »,
[34] Hirst, D. An analytical methodology for
acousmatic music, 2004 Universitat Pompeu
[35] Moorer, J. A. 1975. "On the Segmentation and
Analysis of Continuous Musical Sound by
Digital Computer". Ph.D. thesis, Stanford
[36] S. McAdams - A. Bregman "Hearing Musical
Streams". Computer Music Journal, 1979.
[37] Gygi, B. "Factors in the Identification of
Environmental Sounds". PhD Dissertation:
Indiana University, 2001
[38] Howard, J. H. & Ballas, J. A. "Syntactic and
semantic factors in the classification of
nonspeech transient patterns". Perception &
Psychophysics 28(5), 431-439, 1980.
[39] Bregman, A. S. (1999). Auditory Scene
Analysis: The Perceptual Organization of
Sound. (Second MIT Press Paperback edition)
Cambridge, MA: MIT Press.
[40] Ricard, J. Towards computational
morphological description of sound,
DEA pre-thesis research work, Universitat
Pompeu Fabra, Barcelona, September 2004.
[41] Cambareri G., Sfriso M., Spolaor E.
Segmentazione e classificazione di un file
audio, internal work, Corso di Laurea
specialistica in Ingegneria Informatica (nuovo
ordinamento) Informatica Musicale, Università
di Padova, Dipartimento di Ingegneria
dell'Informazione, 2005.
[42] L.Wyse and M.Slaney, "Toward content based-
audio indexing and retrival and a new apeaker
discrimination tecnique", Inst. Syst. Sci., Nat.
Univ. Singapore,,
Dec 1995.
[43] Kimber, D. – Wilcox L. "Acoustics
segmentation for audio browsers", in
Proc.Interface Conf., Sydney, Australia, July
[44] Pfeifer S., Fischer S., Effelsberg W.,
"Automatic audio content analysis", Praktische
Informtik IV, Univ.Mannheim, Germany,
[45] Marc Leman. Foundations of Musicology as
Content Processing Science. Journal of Music
and Meaning 1 (2003)
[46] Klapuri, A., "Signal Processing Methods for
the Automatic Transcription of Music",
Tampere University of Technology
Publications 460, ISBN 952-15-1147-8, ISSN
1459-2045, 2004.
[47] Martin, K.D., Sound-Source Recognition: A
Theory and Computational Model, PhD thesis,
Massachusetts Institute of Technology, 2000.
[48] Nucibella, F. – Porcelluzzi, S. – Zattra, L.,
"Computer music analysis via a
multidisciplinary approach", here.
... Сайт конкурсу - 54 В середині 2010-х років проєкт «Onuka» (співавтори -Наталія Жижченко і Євген Філатов) здобуває успіх, поєднуючи характерні риси танцювальної електронної музики зі звучанням українських народних інструментів, таких як сопілка, цимбали, трембіти та інші. ...
... Мал.54. Нерозбірливі ноти Мал. ...
Full-text available
The experience of applying technologies for making music is methodized in this textbook. The textbook contains information about history of audiorecording technologies and electronic music instruments and their influence on a music art development as well as emergence of modern music genres. The practical chapter is devoted to the work with the musical software, particularly score writers, audio processing, making phonograms via producing MIDI file and its following processing in DAW and VST instruments. The textbook is designed for music high school students, professional musicians and everyone who is interested in using computer for making music.
... A análise computacional […] poderia aplicar seus resultados atuais na extração automática de características, classificação e segmentação do sons para a análise musicológica, sua representação e transcrição. Como a tarefa de percepção é complexa e subjetiva, o computador pode ser de grande interesse, permitindo alcançar uma objetividade maior através da modelagem (ZATTRA, 2005) Levando em consideração, portanto, o surgimento durante o século XX de tendências estéticas na valorização do som em relação à nota musical, a limitação oferecida pela análise de fenômenos sonoros observados exclusivamente pela partitura, e o crescente campo de estudos de MIR através da análise computacional, este trabalho propõe fazer uso ferramental de descritores de áudio para a análise da obra. ...
... (2) visual analysis -based on representations such as sonograms, features plots and other graphic forms, holding, more or less overtly, the idea that visual representations can reveal to us essential aspects of sounds that would not be explicit or obvious to the raw listening perception; (3) data-driven analysis -which exploits computational tools and techniques, often imported from several different areas, to generate quantitative data and suitable processing mechanisms that may lead to meaningful analytical deductions (EMMERSON, 2016;ZATTRA, 2005). ...
Conference Paper
Full-text available
The text discusses the application of feature-data retrieved and processed by audio descriptors and multivariate statistical methods-particularly, by employing principal component analysis (PCA). Adopting Trevor Wishart's Imago (2002) as a case study, we present an analytical strategy that seeks to correlate the quantitative data and techniques of descriptors and multivariate statistics with qualitative approaches based on listening (solfège, in Schaefferian terms). Finally, we seek to demonstrate the potential assets of localized, precise, dialectical and contextual explorations of these specific techniques to support computer-assisted analytical researches that require the coordination of listening-based and data-informed investigative approaches.
Conference Paper
Apresentamos uma autoanálise dos processos colaborativos (DONIN, 2015; DONIN, FÉRON; 2012) de Nácar (2020), para cello, live-electronics e visuais generativos, e da peça audiovisual Vólpora (2021), realizados à distância e online pelos autores. As criações tiveram como ponto de partida a evocação de imagens poéticas, contudo as decisões práticas dos processos foram distintas e possibilitaram diferentes resultados musicais e criativos.
Full-text available
Noise music and noise itself are puzzling subjects for they consist of numerous definitions and properties. Thus, it is arduous to analyse and interpret the musical pieces in the genre. In this paper, the descriptions of noise and the historical background of Japanoise and noise music are briefly mentioned. Then, analyses were made on several Japanese noise music pieces, including the pieces by musicians and bands Merzbow, Joji Yuasa, Aube, Hijokaidan and Boredoms, by using listening methods and spectrogram analysis.
Full-text available
Neste texto, investigamos como estruturas microtonais foram organizadas por compositores em obras eletroacústicas. Mais que uma análise musical, nos interessa explorar os problemas composicionais emergentes quando nos deparamos com os conceitos e terminologias de classificação de alturas (microtonalidade, temperamento, timbre etc.). Exploramos, portanto, como é e como foi aplicado o parâmetro da altura dentro de obras do repertório, mais especificamente suas qualidades e potenciais dentro da temática da microtonalidade. Para isso, baseamos nossas análises nos estudos de Franck Jedrzejewski, Peter Manning, Curtis Roads, Harry Partch, entre outros. Como resultado, buscamos incentivar a discussão e produção artística acerca do tema, destacando o ofício do compositor em ambientes com suporte tecnológico digital.
This article focuses on the analysis of computer music, that is, music which uses programming languages so that what the listener hears is the result of computer code. One key point in this article is that this music exists with some writing, that is, the computer code. I note that this key point has not been addressed in the latest theories for analysing computer music. Indeed, we often see this music as part of the electroacoustic field, where the audio signal is essential, and where we usually read that those musics are non-written music. After an introduction on this topic, in the second section I will make a distinction between ‘before the signal’ and ‘from the signal’ to organise the theories to analyse electroacoustic music. In the third section, I will focus on computer music and I will show the historical difficulty in considering ‘code’ in musical analysis, mainly with an important exchange between two pioneers, Marco Stroppa and Jean-Claude Risset. In the fourth section I will explain with Jean-Claude Risset and Horacio Vaggione the specificity of computer music: this music is written. Finally, I will look into a recent analysis theory, the Interactive Aural Analysis by Michael Clarke, which seems to fit with the latter specificity.
Full-text available
cote interne IRCAM: McAdams79a
Full-text available
La recherche vise à tracer un parcours comparatif sur le plan esthétique et historique entre la production musicale du CSC de Padoue (Centro di Sonologia Computazionale dell'Università di Padova) et de l'IRCAM de Paris (Institut de Recherche et Coordination Acoustique/Musique) pendant les années 1975-1985. Dans la première partie l'ambiance culturelle et les étapes de la recherche technologique et de la production musicale des deux centres sont reconstruits. La deuxième partie aborde le problème de l'analyse de la musique informatique. Les méthodologies actuelles peuvent être définies d'un côté esthésiques, de l'autre poi͏̈étiques (considérant la définition de J. J. Nattiez). Partant de l'analyse des six œuvres produites au CSC et à l'Ircam (M. Graziani, Winter leaves, W. Motz, Sotto pressione, F. Razzi, Progetto secondo, J. Harvey, Mortuos plango, vivos voco, J. C. Risset, Inharmonique), la thèse cherche à réunir ces deux approches dans une méthode cohérente. Le but est de parvenir à une proposition de méthodologie qui prétend de la part de l'analyste une connaissance profonde des logiciels employés, associée au regard détaché d'auditeur. Les oeuvres analysées utilisent les programmes MUSIC pour la synthèse sonore. Ce trait commun permet de définir des critères de comparaison pour juger les différents approches à la musique par ordinateur.
Sommeil, from Pierre Henry's Variations pour une porte et un soupir, provides the basis for an investigation into listening strategies. Methodological problems raised by the analysis of acousmatic music are considered, followed by a discussion of the three types of listening behaviour revealed in a study of eight subjects’ responses to the piece: taxonomic listening, empathic listening, and figurativisation. Three further, complementary listening behaviours are also considered, and the implications for music analysis outlined.
‘Signed Listening’ is a project that was initiated by Ircam in the spring of 2003. Its goal is to develop computer tools that permit an expanded listening – whether it be for musical analysis, for composition or simply for its own sake with no specific goal in mind (via stereo or computer). This article will briefly present some hypotheses and objectives of the project, and how it touches issues relating, in particular, to electroacoustic music, as well as presenting examples taken from the early stages of our research.