Content uploaded by Petri Toiviainen
Author content
All content in this area was uploaded by Petri Toiviainen on Jul 23, 2017
Content may be subject to copyright.
P1: SFK/SFV P2: SFK
9780521868617c12.xml CUUK1005B-Collins July 24, 2007 15:0
12 The psychology of electronic music
petri toiviainen
Psychology provides an important base from which to understand music,
and is very relevant for electronic music in particular, where psychological
theories have even inspired new compositional explorations. Furthermore,
in analysing and composing electronic music, traditional music theory is
often not applicable. There is no conventional score available on which the
analysis of the music could be based, for the music does not rest solely on
certain standard notated pitch structures and rhythmic frameworks, but
encompasses timbre, spatialisation and other general auditory parameters.
An appreciation of the role of aural cognition is vital for a true engagement
with this field, where any sounding object is fair game.
The purpose of this chapter is to provide an introduction to perceptual
and cognitive processes of music that are fundamental for understanding
electronic music. The chapter begins with a discussion of the neuroscientific
basis of the auditory system. This is followed by a discussion of low-level
phenomena of audition, including the localisation of sound sources, mask-
ing, auditory stream segregation and the perception of timbre. Next, the
perception of pitch is tackled, with a discussion on its relation to alternative
tunings. Finally, basic notions of rhythm perception are introduced. For
each of these parts, electronic music examples illustrating the perceptual
principles will be given. Any and all principles expounded in this chapter
might be taken up and profitably investigated by electronic musicians.
The neuroscientific basis of audition
The auditory system can be partitioned into three processing stages (Pickles
1988; Moore 1997). These are the auditory periphery, the auditory pathway
and the auditory cortex. The auditory peripher y consists of the outer, middle
and inner ear, the auditory pathway of the tracts connecting the ear and the
auditory cortex, and the cortical level primarily of the temporal lobes at the
left and right sides of the brain.
The ear transforms the mechanical energy of sound vibrations into nerve
impulses. The outer ear consists of the pinna and the ear canal, whose
function is to collect and amplify the energy of the sound and lead them to
the tympanic membrane. The ear canal also acts as a closed tube resonator
[218]
P1: SFK/SFV P2: SFK
9780521868617c12.xml CUUK1005B-Collins July 24, 2007 15:0
219 The psychology of electronic music
and amplifies frequencies in the range of 2–5 kHz. The oscillations of the
tympanic membrane are led to the inner ear via three bones (ossicles) of
the middle ear, the malleus (hammer), incus (anvil) and stapes (stirrup).
The vibrations of the ossicles enter the cochlea, a spiral structure of the inner
ear, through the oval window.
The cochlea performs a transformation of the mechanical vibration into
electrical impulses. This is carried out through the movement of the basilar
membrane, which bends rows of hair cells beneath it. The bending of the
hair cells gives rise to electric impulses that encode information about the
periodicity and intensity of the sound.
Due to its mechanical properties, the basilar membrane acts as a fre-
quency analyser. More specifically, the stiffness of the membrane varies
along its length, causing the front end (base) of the membrane to resonate
with higher frequencies, and its rear end (apex) with lower frequencies.
The auditory pathway leads neural impulses from the cochlea to the
cortex. It also contains nuclei that carry out preliminary analysis of the
sound signal with regard to, for instance, its intensity and spatial origin.
The cochlear nucleus sharpens the frequency information contained in the
neural signals. The inferior colliculus plays a role in sound source locating.
The thalamus is considered as the ‘gateway to cortex’ and a ‘gatekeeper
of conscious experience’ (Llinas et al. 1998); all neural information to the
cortex passes through the thalamus.
The auditory cortex is one of the most folded parts of the brain. It is
this part of the auditory system where identification and segregation of
auditory objects occurs; memory-based sound processing also takes place
in the cortex. Although the left and right cortices are mostly similar, some
functional differences have been observed. In particular, the left hemisphere
has been found to be dominant in rhythmic processing, whereas the right
one is more relevant for pitch processing (Zatorre 2003).
There is still much unknown about the functioning of the cortex, but
research in this area is active, and there will certainly be many implications
of this research for musicians in the future. For instance, revealing the neural
determinants of musical emotions is useful for understanding the elements
of music that affect listeners’ mood. Despite the present incomplete knowl-
edge on the functioning of the brain, there have been several applications
to use the activity on the cortex to produce music. In particular, brainwaves
measured with electroencephalogram (EEG), have been used to controlelec-
tronic synthesisers. Pioneers in this field include Richard Teitelbaum and
David Rosenboom. For instance, in his work In Tune (1967), Teitelbaum
combined amplified EEG signals with sounds of heartbeat and breath. In
his composition Ecology of the Skin (1970), Rosenboom used ten live EEG
performers to interactively generate immersive sonic environments.
P1: SFK/SFV P2: SFK
9780521868617c12.xml CUUK1005B-Collins July 24, 2007 15:0
220 Petri Toiviainen
Figure 12.1 (a) Schematic presentation of the cochlea; (b) Excitation pattern of basilar membrane
for high, medium, and low frequencies; (c) Auditory pathway; 1. Auditory nerve; 2. Cochlear
nucleus; 3. Superior olive; 4. Lateral lemniscus; 5. Inferior colliculus; 6. Thalamus; 7. Auditory
cortex
Several computational models of the auditory processing have been pro-
posed and implemented as computer algorithms. For instance, the IPEM
Toolbox (Leman, Lesaffre and Tanghe 2001) contains an implementation of
models for pitch, sensory dissonance, onset detection, beat and metre, and
timbre characteristics. Such models could have various applications as, for
instance, artificial ears that could be used by composers to make perceptual
analysis of their music. It must be noted, however, that a well-grounded
theory of auditory processing only exists for the peripheral level, whereas
models of the subsequent levels are much more speculative.
Localisation
Spatialisation has played an important role in electronic music throughout
its history. Humans have a remarkable ability to localise sound sources accu-
rately and rapidly. Sound localisation can be divided into three components:
localisation of azimuth, elevation and distance. In what follows, each of these
components is discussed in turn.
Localisation of azimuth refers to specifying the direction of the sound
source on the horizontal plane. The two main cues used in this process
are the Interaural Time Difference (ITD), and the Interaural Level Difference
(ILD). The ITD is caused by the difference in the time it takes for the sound
wave to reach the two ears. This is illustrated in Fig. 12.2a. The ILD, on the
other hand, is caused by the fact that the ear that is more distant from the
sound source receives less sound energy due to the head’s shadow. This is
demonstrated in Fig. 12.2b. To determine the azimuth, the auditory system
uses both ITD and ILD information.
P1: SFK/SFV P2: SFK
9780521868617c12.xml CUUK1005B-Collins July 24, 2007 15:0
221 The psychology of electronic music
rsin θ
r
θ
rθ
d = r θ + rsin θ
ab
Figure 12.2 (a) Interaural Time Difference. The distance from the sound source (black circle) to
the right ear is d=r+rsinlongerthantotheleftear,whererdenotestheradiusofthehead
and the azimuth of the sound source. (b) Interaural Level Difference. Head shadowing reduces
theintensityofthesoundarrivingtotherightear
The localisation of elevation of a sound source is less accurate than that
of azimuth. Still, we can easily tell whether the sound source is ahead, above
or behind us. In this process, there are no binaural cues (ITD, ILD) to rely
on; rather, the most important cues are related to the spectral shape of the
perceived sound. In particular, the received sound spectrum is modified by
reflections from the pinna that depend on elevation.
For localisation of distance, the loudness of a sound source is an evident
cue, but cognitive knowledge about the quality of the particular sound has
to be applied to utilise this cue. For instance, shouting from a long distance
can have a higher perceived loudness than whispering from a short distance.
A further cue is motion parallax, which refers to the fact that translational
movement of the listener causes larger azimuth change for nearby objects
than for distant ones. Further cues, sometimes used for musical purposes, are
the loudness ratio between the direct and the reverberant sound (Zahorik
2002), and the brightness of timbre. More specifically, high intensity of
reverberation gives an impression of a distant sound source. Furthermore,
a dark (low-pass filtered) timbre may give an impression of a distant source,
because high frequencies attenuate faster than low ones in the air.
One of the first electroacoustic compositions, Edgard Var`
ese’s Po`eme
´electronique, was presented at the Brussels World’s Fair in 1958. The audio
part of this multimedia composition consisted of a three-track tape record-
ing, each track of which was distributed dynamically to 425 speakers through
an eleven-channel sound system with twenty amplifier channels. Karlheinz
P1: SFK/SFV P2: SFK
9780521868617c12.xml CUUK1005B-Collins July 24, 2007 15:0
222 Petri Toiviainen
Stockhausen has used spatial sound as an integral part of his work. He
produced the first true quadraphonic composition for electronic sounds,
Kontakte (1960). In this work, Stockhausen used a turntable system with a
rotating loudspeaker mechanism, recording to four-channel tape via four
microphones spaced around the table, creating an effect of sounds orbiting
around the audience.
With methods of digital signal processing it is possible to spatialise
sounds accurately by simulating the ITDs and ILDs as well as filtering and
adding reverberation. John Chowning (1971) described techniques for the
simulation of moving sound sources that are based on the Doppler effect
as well as reverberation effects. The Doppler effect refers to the change of
frequency caused by the movement of the sound source relative to the lis-
tener, and can be used compositionally to create an illusion of movement.
Chowning’s composition Turenas (1972) is a quadraphonic work based on
these techniques. In the domain of digital signal processing, various tech-
niques of sound spatialisation have been developed since and applied by
composers of electroacoustic music. These include ambisonics, holophon-
ics, wave-field synthesis, Dolby 5.1 Surround, and Digital Theater Systems
(DTS) as well as binaural systems based on head-related transfer function
(HRTF). Today’s efficient computers and sound-processing software allow
relatively easy experimentation with sound spatialisation effects. This issue
is further discussed in chapter 13 of this book by Natasha Barrett.
Masking
Masking refers to a phenomenon whereby a signal with a low intensity (the
maskee) is made inaudible by a stronger signal (the masker). There are two
types of masking: simultaneous and temporal. In simultaneous masking, the
strength of masking depends on the frequency content of the two signals. For
instance, with pure tones, the masking effect is stronger the closer the fre-
quency of the maskee is to that of the masker. Moreover, masking is stronger
for frequencies above that of the masker than below. Simultaneous masking
is caused by the overlap of excitation patterns on the basilar membrane. The
range of frequencies within which masking occurs for a given frequency is
referred to as the critical bandwidth. This bandwidth is about 90 Hz wide
for sounds below 200 Hz, and increases to about 900 Hz for frequencies
around 5000 Hz. The degree to which complex sounds mask other sounds,
or are masked by other sounds, depends, in addition to their intensity, on
their spectral content. For instance, a higher intensity difference is needed
for a sinusoidal masker to mask a noise-like maskee than is needed for a
noise-like masker to mask a sinusoidal maskee.
P1: SFK/SFV P2: SFK
9780521868617c12.xml CUUK1005B-Collins July 24, 2007 15:0
223 The psychology of electronic music
Temporal masking refers to the phenomenon whereby a soft tone is
masked by a louder tone that occurs shortly before (post-masking) or after
(pre-masking) the soft tone. For a constant difference in the intensities of
the masker and the maskee, post-masking has been found to have a longer
range than pre-masking. Typically, post-masking occurs within about 50–
200 msec (milliseconds) after the removal of the masker, whereas the range
of pre-masking is about one tenth of this (Zwislocki 1978). All three forms
of masking are utilised in Perceptual Audio Coding, such as MP3, MiniDisk,
and Ogg Vorbis.
The phenomenon of masking has a number of implications to compo-
sitional practice. For instance, a loud sound object in a composition may
make weaker sound objects in nearby frequencies inaudible; assigning the
sound objects different frequency ranges will diminish the effect of mask-
ing. For instance, a melodic line is often played in a higher register than
the accompaniment to make it better audible. The implication of temporal
masking is that the audibility of less loud parts of the musical material can
be improved by placing them at different temporal locations than louder
parts. Such techniques are applied by many musicians and mix engineers in
their practice.
Auditory streaming
Our auditory system has a remarkable capacity to make sense of the sounds
we receive from our environment. In particular, it can extract from the
sound signals we receive meaningful chunks of information that correspond
to real-world activities. For instance, in a room full of people talking to
each other we can usually without any trouble concentrate on the speech
of a single person. In other words, our perceptual system is capable of
extracting meaningful auditory streams from the perceived sound signals.
When listening to music, we also tend to hear the auditory information as
a collection of streams, such as parts in counterpoint, melodic lines, inner
voices, bass lines, and accompaniment.
The research on the formation of auditory streams has a long history.
The founders of Gestalt psychology, such as Paul Ehrenfels (1890) and Max
Wertheimer (1923), initially presented musical examples to support their
notions. The main ideas of Gestalt psychology can be summarised into a
few principles. The ones that are most relevant with regard to auditory
streaming are the following:
Principle of proximity: objects that are close to each other tend to be
grouped together
P1: SFK/SFV P2: SFK
9780521868617c12.xml CUUK1005B-Collins July 24, 2007 15:0
224 Petri Toiviainen
Principle of similarity: objects that share similar characteristics tend to be
grouped together
Principle of good continuation: there is a preference for continuous forms
Principle of closure: objects that seem to form closed entities tend to be
grouped together
Principle of common fate: objects that move together tend to be grouped
together
In addition to the Gestalt psychologists, the problem of auditory stream-
ing has been studied by researchers such as Hermann von Helmholtz, Carl
Stumpf, Jay Dowling, Diana Deutsch, Leon van Noorden, David Wessel and
Stephen McAdams. The most influential work in this field is the book Audi-
tory Scene Analysis by Al Bregman (1990). In this book, Bregman presents
detailed accounts of various processes involved in auditory streaming, many
of which are based on the notions originally presented by the Gestalt psy-
chologists. Bregman distinguishes between two main types of processes
that are involved in auditory streaming. These are sequential integration
and spectral integration.
Sequential integration refers to the putting together of events that follow
one another in time. Musical events can coalesce into a single stream if they
are sufficiently proximal in time and/or pitch. Moreover, the closer in time
two events are, the more proximal in pitch they have to be in order to be
integrated into a single stream. An example of this dependence is graphi-
cally depicted in Fig. 12.3a. The dependence between pitch and temporal
proximity in stream formation was quantified by van Noorden (1977), who
defined the temporal coherence and fission boundaries for auditory stream-
ing. An example of auditory stream formation can be found, for instance, in
Andean pipe music where musicians play alternate notes that, due to their
pitch proximity, are perceived as a single stream. Sequential integration can
also be based on similarity in timbre or loudness, so for instance, if a musical
passage contains tones played by two instruments, these tend to be heard as
two separate streams.
Spectral integration refers to integrating components that occur at the
same time in different parts of the spectrum. There are a number of princi-
ples that govern this phenomenon. First, we tend to group frequency com-
ponents by harmonicity. More specifically, components sharing the same
fundamental are likely to come from the same source, and are grouped
together. Second, we group frequency components by onset. This means
that frequency components that have proximal onset times are likely to
come from the same source, and are grouped together. Finally, we group
frequency components based on similarity of their temporal evolution.
For instance, spectral components sharing the same frequency modulation
P1: SFK/SFV P2: SFK
9780521868617c12.xml CUUK1005B-Collins July 24, 2007 15:0
225 The psychology of electronic music
Figure 12.3 (a) Grouping by proximity. A slow sequence of tones with alternating frequencies
(left) is perceived as a single stream; the same sequence played twice as fast is perceived as two
separatestreams;(b)Groupingbycommonfate.Acollectionofpartialswithnofrequency
modulation is perceived as a single tone (left); when frequency modulation is introduced, tones
with similar FM pattern are grouped together, resulting in a percept of three separate tones
(right); (c) Principle of ‘old-plus-new heuristic’. Two frequency slides are perceived as separate
tones (left); when a tone burst is played between the slides, they are perceived as a single tone
pattern (such as vibrato) are likely to come from the same source, and are
grouped together (Cook 1999). An example of this phenomenon is shown
in Fig. 12.3b. The last two principles are instances of the Gestalt principle
of common fate.
A further principle involved in spectral integration is what Bregman
refers to as the ‘old-plus-new heuristic’. This refers to perceptual continuation
of an old sound at the presentation of a more complex sound. In other words,
if part of a present sound can be interpreted as being a continuation of a
previous sound, the auditory system tends to make this interpretation. An
example of this principle is illustrated in Fig. 12.3c.
Perception of auditory streams plays a crucial role in the parsing of
musical compositions. Based on the principles described above we hear,
for instance, separate voices in a musical work. In much electronic music
we cannot talk about melodic lines or even pitch, but the music still has
temporal and spectral dimensions, to which the streaming principles apply.
P1: SFK/SFV P2: SFK
9780521868617c12.xml CUUK1005B-Collins July 24, 2007 15:0
226 Petri Toiviainen
As a result, in the total mass of sound we perceive layers that are independent
from each other and that at some points fuse into a single percept.
Similarities based on timbre can be used to compose streams. On the
other hand, introducing timbral differences within a sequential stream may
add interest. This has been used, for instance in Klangfarbenmelodien by
Schoenberg and Webern. In a melodic line, subsequent tones are played
with different instruments. The rapid changes in timbre interfere with the
smooth sequential integration, thus bringing interest to the music.
Ver y oftena clear segregation of two streams in music is desired. To obtain
this, one can, for instance, introduce differences between the spectral con-
tent of the two streams. A classical example of this is the singer’s formant, a
spectral bulge around the frequency of 3 kHz that helps the singer to be dis-
tinguished form the ensemble. Furthermore, the principle of common fate
can be applied by introducing individual vibrato patterns to the streams. A
further means is to add minor temporal deviations to otherwise simultane-
ous events. Even a slight asynchrony renders the two streams perceptually
separate. Sound spatialisation provides a further means to separate streams:
sound sources are segregated more easily when they are placed at different
directions in the sound field than when they appear to come from the same
direction.
Timbre perception
The timbre of sound is a complex phenomenon. There have been a number
of studies aiming at extracting the most salient acoustic attributes affecting
the perception of timbre (most aimed at studying monophonic instrument
tone colour, rather than general sound objects). A widely used method
is similarity rating (SR): subjects are asked to rate, on a given scale, the
dissimilarity of all possible pairs in the set of stimuli. Multidimensional
scaling (MDS) is then used to map the tones onto a low-dimensional space –
frequently referred to as the timbre space. Most of these studies have found
that a three-dimensional timbre space represents the dissimilarity ratings
to a sufficient degree of precision (e.g. Grey 1977; McAdams et al. 1995).
The first dimension in the MDS solution has been found to correlate with
the spectral centroid, which corresponds to the perceived brightness of a
tone. The second dimension in the MDS solution relates to the attack time,
that is, the time it takes for the amplitude of the tone to reach the maximal
value. As regards the third dimension, there is more discrepancy between the
studies. It has been associated with spectral variation over time or spectral
irregularity.
Most of the work on timbre has concentrated on single musical tones,
that is, monophonic timbre, whereas the overall timbre of a musical piece,
P1: SFK/SFV P2: SFK
9780521868617c12.xml CUUK1005B-Collins July 24, 2007 15:0
227 The psychology of electronic music
also referred to as the polyphonic timbre, has received less attention. Recently,
however, there has been increased interest in the study of this phenomenon
(Aucouturier 2006). Much of this work has been carried out in the field of
Music Information Retrieval, where it has been found that various descrip-
tors of polyphonic timbre can be leveraged, for instance, for the automatic
classification of audio. A practical application of polyphonic timbre analysis
is the Shazam query-by-example system for recognition of music via mobile
phone. The system is based on a method of audio fingerprinting; relative
positions of peaks in the spectrum of the audio query are located and con-
verted into a fingerprint, which is then matched with the fingerprints of the
pieces in the database.
A wider notion of timbre has played a vital role in electroacoustic music.
For instance, the French composer Pierre Schaeffer (1910–95), the inventor
of musique concr`ete, started from concrete sounds such as voices, noises,
as well as sounds of prepared and conventional instruments, experimented
with them, and abstracted them into musical compositions. Examples of
such works by him are Suite pour quatorze instruments (1949), which is based
on timbral transformations of orchestral instruments, and Symphonie pour
un homme seul (1951), co-composed by Pierre Henry, which employed,
among other things, the sounds of the human body. The composition Atmo-
sph`eres (1961) by the Hungarian composer Gy¨
orgy Ligeti is a classic example
of timbral composition. Written for a large orchestra, it abandons the con-
cepts of melody, harmony and rhythm, while concentrating solely on the
timbre of the sound produced. Atmosph`eres opens with a large cluster chord
comprising every tone in the chromatic scale over a range of five octaves.
Because of the pitch proximity of adjacent components in the cluster, the
auditory systems cannot resolve every tone. Therefore, what is perceived is
the timbre of the sound mass rather than a chord.
Advanced methods of digital signal processing allow operations on the
microscopic level of sound structure, dubbed microsound by Curtis Roads
(2001). Granular synthesis uses small pieces of sound, typically with a
length between 1 and 50 msec, to build sound textures, or clouds. The
first composer to use this technique was Iannis Xenakis in his composi-
tions Analogique A et B (1958–9). Extended notions of timbre and sound
materials are discussed in the next chapter in particular.
Pitch perception and alternative tunings
Pitch is a perceptual attribute of a tone that depends mainly on its frequency
content. For complex harmonic tones, the perceived pitch is equal to the
fundamental frequency of the complex tone, that is, the frequency of the
first partial. For inharmonic tones, such as a bell tone, the perceived pitch
P1: SFK/SFV P2: SFK
9780521868617c12.xml CUUK1005B-Collins July 24, 2007 15:0
228 Petri Toiviainen
may, however, be unclear. In fact, such tones often elicit a percept of sev-
eral simultaneous pitches. Terhardt et al. (1982) proposed a model of pitch
salience according to which this attribute depends on the degree of har-
monicity of the tone, that is, the degree of coincidence of the subharmonics
of the partials.
The place theory of pitch perception, originally suggested by George von
B´
ek´
esy (1963), states that the pitch percept can be explained by the locus
of maximal vibration on the basilar membrane. The rate theory (Seebeck,
1843), on the other hand, states that the neural firing patterns encode the
periodic structure of auditory stimuli. According to a commonly accepted
view, both place and rate information are used by the auditory system to
determine pitch, with rate information dominating for low frequencies and
place information for high frequencies.
Traditionally, most music is comprised of a collection of discrete pitches,
or scales, instead of a continuum of them. Furthermore, it is common that the
scales repeat themselves after an octave, or a frequency ratio of two. Many
studies have indicated that tones an octave apart are perceived as highly
similar (Burns 1999) – a phenomenon referred to as octave equivalence.
Conventionally, the number of pitches per octave has been between five and
seven (Carterette and Kendall 1999). The repetition of pitch intervals after
each octave is, however, dropped in some scales, such as the Bohlen–Pierce
scale (see below). Moreover Iannis Xenakis uses non-octave scales in his
works, such as Tetora for string quartet (1990), and discusses them in his
writings on sieve structures (Xenakis 1992).
Scales are derived from tuning systems. In particular, a scale is a subset of
a tuning system, often uneven in the sense that it consists of pitch intervals of
varying sizes. There is a long tradition in the development of various tuning
systems in Western culture, starting from Pythagoras (582–507 BC), who
observed that consonant musical intervals produced by a vibrating string
were associated with simple integer ratios of string length. In the Western
world, the 12 Tone Equal Tempered tuning (or 12TET tuning) is the most
prevalent nowadays. Although it is the most studied tuning system from
both music-theoretical and perceptual points of view, it must be noted that
that it is not universally accepted.
The 12 Tone Equal Tempered tuning consists of intervals of equal size
with a frequency ratio of 21/12 between subsequent tones. It provides an
approximation of tunings based on simple frequency ratios, while allowing
for modulation between keys. It is possible to construct equally tempered
scales beyond the 12TET tuning by using a frequency ratio of 21/k,wherek
can be any integer. Many of these alternative tunings have been proposed to
provide an approximation for a tuning based on simple frequency ratios. For
instance, the 19TET tuning was used by the composer Guillaume Costeley
as early as the sixteenth century. This tuning serves as an approximation for
P1: SFK/SFV P2: SFK
9780521868617c12.xml CUUK1005B-Collins July 24, 2007 15:0
229 The psychology of electronic music
mean-tone temperament. The 22TET tuning, proposed by the nineteenth
century English scholar R. H. M. Bosanquet, provides an approximation of
the five-limit just intonation, that is, a tuning consisting of frequency rations
that can be expressed with the primes 2, 3, and 5. Many composers, such
as Charles Ives and Krzysztof Penderecki, have composed music using the
24TET tuning, consisting of half-semitone intervals. Higher-order equal
temperaments that have been proposed include the 31TET, 53TET, and
72TET tuning.
One great advantage of computers is the ease with which they allow
exploration of alternative tuning systems. The 12TET tuning was originally
adopted in order to allow flexible modulations between keys, for which the
just intonations were not suitable. Today’s computer technology allows the
use of adaptive tunings, in which the just intonation is adapted to fit with
the current key (see, e.g., Sethares 1994, 1999).
It is not necessary that an equally tempered tuning be based on the octave.
Generally, one can design equally tempered tunings using a frequency ratio
of p1/k, where p can be any number. For instance, the Bohlen–Pierce tun-
ing is a 13TET tuning based on a tritave,orafrequencyratioofp=3.
This tuning can be seen as an approximation of a just intonation system
based only on ratios of odd whole numbers, therefore being appropriate for
timbres containing only odd harmonics. Composers who have utilised the
Bohlen-Pierce scale in their works include Charles Carpenter (e.g., Frog `a
la Pˆeche, 1994), Juan Reyes (e.g., ppP, 1999–2000), and Richard Boulanger
(e.g., Solemn Song for Evening, 1990).
Rhythm perception
The auditory system can accurately detect the temporal structure of the
sounds it receives. In the shortest timescale, there are certain thresholds that
are useful for electronic musicians. In binaural hearing, temporal differences
as short as twenty microseconds can be detected. For monaural hearing, the
threshold of simultaneity for clicks is about two milliseconds and for musical
tones about twenty milliseconds. Echoes are discriminated if their temporal
distance from the direct sound is at least fifty to sixty milliseconds.
Much music is organised so as to contain temporal periodicities that
evoke a percept of regularly occurring pulses,orbeats. The ability to infer
beat and metre from music is one of the basic activities of musical cognition.
It is a rapid process: after having heard only a short fraction of music, we
are able to develop a sense of beat and tap our foot along with it. Even
if the music is rhythmically complex, containing a range of different time
values and probably syncopation, we are capable of inferring the different
periodicities and synchronising to them.
P1: SFK/SFV P2: SFK
9780521868617c12.xml CUUK1005B-Collins July 24, 2007 15:0
230 Petri Toiviainen
Figure 12.4 ResponseofaresonatingoscillatorbanktoanexcerptfromtheScottishfolkmelody
Auld Lang Syne. a) waveform; b) output of an onset detector; c) outputs of resonating oscillators;
d) summed output of all oscillators. Notice the form of the summed output indicating a
hierarchical structure of beat strengths.
A rhythmical sequence usually evokes a number of different pulse sen-
sations, each of which has a different perceptual salience. The salience of a
given pulse sensation depends on a number of factors related to the surface
and structural properties of music. These factors include the frequency of
tone onsets that coincide with the pulse (Palmer and Krumhansl 1990),
and the phenomenal accents of these notes (Lerdahl and Jackendoff 1983).
Phenomenal accents arise from changes in surface properties of music such
as pitch, duration, loudness and timbre. For instance, a long note is usually
perceived as more accented than a short one (Parncutt 1994).
A further factor that affects the salience of pulse sensation is the pulse
period. According to a number of studies (Fraisse 1982; Parncutt 1994;
van Noorden and Moelants 1999), the most salient pulse sensations have
a period of approximately 600 msec, the region of greatest salience being
between 400 and 900 msec. One should note that this range of periods
corresponds roughly to the speed of some basic human activities such as
heartbeat, locomotion, and infant sucking.
In much Western music, the perceived pulses are often hierarchically
organised, and consist of at least two simultaneous levels, whose periods
have an integer ratio. This gives rise to a percept of regularly alternating
strong and weak beats, a phenomenon referred to as metre (Cooper and
Meyer 1960; Lerdahl and Jackendoff 1983). In Western music, the ratio
of the pulse lengths is usually limited to 1:2 (duple metre) and 1:3 (triple
metre). It must be noted, however, that this kind of hierarchical organisation
of pulses does not exist in all music. Examples of types of music that do not
P1: SFK/SFV P2: SFK
9780521868617c12.xml CUUK1005B-Collins July 24, 2007 15:0
231 The psychology of electronic music
possess such metrical structure are found in Norwegian Hardanger fiddle
music, Lappish yoiks, West African polyrhythmic percussion music, and
Eastern European aksak dance music. Electronic music allows the possibility
to explore non-standard and alternative metrical structures which diverge
from Western common practice to a wider sphere of human music-making,
whilst still grounded in psychological percepts.
The perception of pulse and metre has been modelled computationally
using a variety of different approaches (Gouyon and Dixon 2005; Collins
2006; also see chapter 10). Fig. 12.4 displays the response of a model of
resonating oscillators to an excerpt of music. Whenever a note onset occurs,
oscillators that are in synchrony with it are excited (Large and Kolen 1994;
Toiviainen 1998).
Conclusion
This chapter has reviewed some of the important aspects of music per-
ception and cognition that can be regarded as useful for gaining better
understanding of the perception of electronic music. Although the different
musical elements have been discussed separately, their interaction may also
play a role in music perception, though this aspect has been less studied.
Most of the processes discussed in this chapter are relatively low-level
ones. Higher-level aspects of cognition that are relevant include the percep-
tion of form and the effect of memory. In general, these processes are less
well understood than the low-level ones. This is due to the increased effect of
individual and cultural background on these processes and the consequent
larger inter-individual variation.
Obviously, due to space limitations, many of these aspects have been
described on a rather superficial level. To gain a better understanding of these
processes, the interested reader is directed to works such as Deutsch (1999),
Snyder (2000), and McAdams and Bigand (1993). Perceptual principles
discussed in this chapter are intimately related to many issues that arise in
other chapters of this companion.
Electronic music provides a range of devices for the production of sound
and musical material that is far more extensive than that available with more
traditional instruments. While these tools facilitate versatile expression of
musical ideas, they also make it possible to produce musical material that
exceeds the capacity of the human perceptual system. Therefore, for an
electronic musician, being aware of the capabilities and limitations of human
auditory processing is crucial for efficient communication of musical ideas
and exploration of new grounds.
