You are on page 1of 9

95

On measuring and interpreting microtiming

MARC-ANTOINE CAMP, LORENZ KILCHENMANN, THOMAS VOLKEN,


OLIVIER SENN

Emotions, physical movements, and interactions of participants in a musical situa-


tion interrelate with microtemporal structures of sounds. When performing musical
phrases and patterns, we arrange sounds along metrical frames, but usually with
considerable microtemporal variability. A jazz drummer may place his ride strokes
before the walking bass; the pianist models the phrases of a 19!h century European
composition by agogic conventions; crooners let their voices flow over the contrast-
ing rhythmic accompaniment. Such microtiming has received considerable attention
by music researchers and has been acoustically described by. the aid of visual sound
representations -1 However, there is a lack of comprehension regarding the perceptual
dimension of musical timing. Despite attempts to formulate psychoacoustic models
for the temporal perception of acoustic stimuli; we still do not know how to deal
with microtemporal variability in combinations of musical sounds with different
acoustical properties, e.g. with short and long onset rise times.
The physical onset of a sound, the moment in which its amplitude envelope starts
rising above the noise floor, does not equal the moment we perceive the sound. If
we align the physical onset of a percussion sound (whose maximum amplitude level
is reached within one millisecond) with the physical onset of a bowed string sound
(having an onset rise time of.several centiseconds), the sounds will not necessarily
be judged perceptually as synchronous. On the contrary, aligning dissimilar sounds
by their physical onsets can lead to the perception of asynchrony. To be able to
quantify the temporal relation of sounds precisely as they are perceived, a range
of questions must be answered, among them the issue of the acoustical conditions
of auditory synchronicity. Answers have been given from several disciplines. Psy-
choacousticians, linguists, neuroscientists, ethnomusicologists, music researchers,
sound engineers, and computer scientists have directly and indirectly contributed to
the complex question of microtemporal perception. This paper provides an overview
of the current research and a report on an experiment with twelve synchronization
tasks, preceded by three conceptual remarks.
96 97

Concepts (Collier and Collier 2002:467). In another study, analyzing rubato in jazz music,
the "attack point" of each tone was detected by the author by ear and eyes at "the
Musical synchronicity amplitude zero-crossing most immediately preceding the beginning of the (quasi-)
Music studies dealing with synchronicity of sounds are outweighed by experiments periodic portion of the tone," taking the average of three measurements some weeks
concerning asynchronies. Researchers have measured differences between tones on apart (Ashley 2002:315). A further jazz study defines the perceptual "tone onset"
the same metrical time (mainly of pianists) ranging from 30 ms to 90 ms.2 It has been as "the moment in time when the amplitude of a tone reached a level approximately
suggested that, in musical performances, acoustical asynchronies between voices 15% below its maximum amplitude" (Rose 1989:6).
function as a means of expression and transparency. Musicians highlight voices These approaches to quantifying microtemporal variability in music are forcedly
or structurally important notes by playing them with higher intensity and slightly pragmatic, since there is no satisfactory model to predict the PMO of a musical
displaced from notes of other voices. As music listeners we assign such acoustically tone (Kopiez 2005:470f.). We only know from previous studies on the p-center of
displaced tones to the same time point in the metrical structure. syllables and the perceptual attack time of musical.sounds that a sound's PMO is
For the purpose of this study, two musical sounds are defined as being synchro- located at a time point somewhere between its physical onset and the maximum of
nous in perception when listeners are not capable of saying whether one starts earlier its amplitude envelope.
than the other. We hypothesize that musical synchronicity can be described as,the
coincidence of two sounds' perceptual moments of occurrence. Validity of experimental design
In dealing with the questions on sounds' PMO, the choice of an appropriate experi-
Perceptual moments of occurrence mental design is decisive in gaining relevant data. In psychoacoustical experiments,
The perceptual moment of occurrence (PMO) of a sound has been defined as its researchers usually split the complex realities of sounds into components to reveal
center and as the relevant point in time of its synchronization with other sounds. The how auditory stimuli are processed. Using sinusoid sounds or noise bursts with linear
concept was developed first in linguistics. The "perceptual center" ("p-center") of amplitude envelopes for listening experiments, single factors influencing the loca-
a syllable (Morton et al. 1976) is not located at the syllable's beginning. Rather, it tion of PMO are controllable. But as these synthetic sounds, isolated from musical
is shifted from the physical onset of the prevocalic consonant toward the following phrases and patterns, are not part of our common auditory and musical experience,
vowel by up to several centiseconds.3 Concerning sung language, it has been shown one may question whether the results of such analytical fractionation can be trans-
that pianists synchronize .their playing to coincide with the beginning' of singers' ferred and applied to complex musical contexts (Bregman 1993; Langner and Goebl
vowels, and not with the prevocalic consonants (Sundberg and Bauer-Huppmann 2003:69; Waadeland 2001:24).
2007). Several factors have been found to determine the time location of the p- For the experiment reported here, real instrumental stimuliwith their non-linear
center, both in speech and in synthetic non-speech sounds.4 amplitude envelopes and rich spectral shapeswere used. Participants, all of whom
Whereas in speech phonetic structure and its acoustical realization influence the had professional musical training, had to synchronize these stimuli with clicks, as-
p-center location (Cooper et al. 1986), musical sounds have to be dealt with in their suming that metronome click trains are used in daily exercise and studio recording
own way. John W. Gordon conducted the only experiments so far on the PMO of and can therefore be considered part of the participants' habitual auditory experi-
musical sounds (Gordon 1987). Gordon (1987:88) used a concept analogous to the ence. Contrary to our claim for closeness to musical reality, we presented tones
p-center, called the "perceptual attack time," which he defined as "the time a tone's outside of a musical context as isolated stimuli, inducing subjects to a constrained
moment of attack or rhythmic emphasis is perceived." analytical focus on single sounds. We thereby followed an experimental design
Gordon's results will be discussed below. Here it is important to note that refer- adopted by Gordon (1987), but we used a smaller amount of stimuli and invited a
ences to Gordon's study and an inter-subjectively founded definition of sounds' larger number of participants. The focus of the current experiment was on the onset
PMOs are mostly absent in studies measuring and analyzing microtemporal struc- rise time of sounds' amplitude envelopes, with the assumption that the synchronizing
tures in music.5 Instead, researchers have chosen diverging methods to define and of a click with an instrumental tone would indicate the PMO of that tone.
measure the onsets of sounds. In a study on timing in Louis Armstrong's solos, "all
onset times were selected to be positive-going zero crossings at the points where
the waveforms began to be discernibly differentiable from the background noise"
98 99

Method Equipment
A MacBook Pro Core 2 and the DAW Motu Digital Performer 6 were used. D/A
Musicians converter was a Metric Halo 2882. Participants heard the stimuli with Sony MDR
41 students from the Lucerne School of Music participated in the study as volun- 7506 headphones.
teers. They were recruited by an email call and paid for their participation (30 Swiss
Francs). One musician showed problems with the experimental procedure; data of Design and procedure
his experiment were not included in the analysis. Out of the 18 male and 22 female A participant sat in a comfortable chair at a quiet office in the Lucerne School of
students, there were six singers, four guitarists, four drummers, seven string players Music. The purpose of the study was explained and participants were assured of the
(violin, viola, violoncello), eight brass players, eight pianists and three other instru- anonymity of their individual results. The participant was asked to fill out a short
mentalists. 27 students were in a Bachelor program, 13 in a Master program. Most questionnaire providing some basic biographical information as a musician, includ-
participants were in a performance program for. European Art Music (23) or for Jazz ing musical preferences both in their own playing, and as music consumers. In each
(11); others were enrolled in the music pedagogy (3) and music theory programs (1). task, participants were instructed to synchronize two sounds, so that these were
The age range was 20 to 35. Participants had an average of 12.6 years' experience heard as musically synchronous. There was no time restriction for task solving, but
on their principal instrument (standard deviation: 4.0 years) and 5.6 years' experi- in the initial explanation participants were alerted to the possible effects of fatigue
ence on a second instrument (standard deviation: 4.5 years).Two musicians reported over time for this special listening activity. To control for sequence effects, subjects
partial hearing loss; as their experimental results were within the standard deviation, were randomly assigned to two different groups, which varied in the sequence of the
their data were included in the set for analysis. synchronization tasks performed. There was one test task, in which participants had
to synchronize a clarinet to a trumpet sound, followed by twelve synchronization
Stimuli tasks used for analysis.
Stimuli for the synchronization takes were a synthetic click and seven samples of in- One member of the research team conducted the experiment, starting every task
strumental tones (Table 1). The samples were taken from Vienna Symphonic Library and keeping a record of the results. Before every task, the participant was told which
Solo Strings (violin sul tasto, viola arco, violoncello arco, contrabass pizzicato), Na- instrumental sounds she or he had to synchronize. The two sounds were presented in
tive Instruments Akoustik Piano (piano), Toontrack DFH Superior (Hi hat cymbals, a loop of 4000 ms, appearing in their initial position with a randomized difference
Ride cymbals). In order to level perceived loudness between all stimuli, the samples of their physical onsets. The participant could shift one of the sounds ("movable
were normalized using the Waves PAZ Analyzer in an A-weighted RMS setting. sound") back and forth in intervals of 20 and 1 ms by pressing the arrow keys on the
computer keyboard. When the participant heard the movable sound as synchronous
with the second one ("fixed sound"), she or he notified the researcher. Musicians had
Tab. 1: Stimuli used for the synchronization tasks no visual aids; some of them closed their eyes for concentration.

Pitch or frequency Peak of amplitude Total stimulus duration


range Hz envelope ms (rounded)
Time after noise floor Results
ms (onset rise time)
lj
Click not determinable 0.07 2 Results are shown in table 2. The average differences between the physical onsets of
Hi hat cymbals 0-22000 5.76 260 the synchronized sounds by all forty musicians range from -6.225 to 46.5 ms with
Ride cymbals 260-22000 1.77 4700 standard deviations ranging from 7.577 to 39.744 ms. The median values within the
Piano 234=b 27.71 2850
Contrabass pizzicato 77 = EHat 37.69 3950 distribution of results range from -7 to 46.5 ms.
Viola arco 441= a' 534.56 2700 The data were evaluated by several factorial repeated-measure analyses of vari-
Violoncello arco 330 = e 810.93 3170
Violin sul tasto 465=b' 1666.21 2630 ance. The sequence of the task had no influence on the results (F=2.84, p=0.09),
so that learning or fatigue effects could be excluded. Half of the participants solved
a single task in less than two minutes. The time spent for solving a task had no
Tab. 2: Results of the twelve synchronizing tasks, showing median and mean difference time between the physical onsets of the sounds
for the forty participants, as well as the standard deviation.
Task Movable sound Fixed sound Difference of the sounds' Standard
physical onset in synchro- deviation ms
nized position ms
Median Mean
I Click Ride 1 0.625 8.995
2 Piano Violin sul tasto 46.5 46.5 39.744
3 Click Viola arco 26.5 25.45 20.283
4 Contrabass pizzicato Piano 4 6 13.199
5 Click Contrabass pizzicato -2.5 -1.3 12.298
6 Click Piano -4.5 -4.275 8.776
7 Ride Contrabass pizzicato -3 -3.9 11.575
8 Click Violoncello arco 29.5 30.975 24.870
9 Hi hat Ride -5 -6.225 7.577
10 Click Violin sul tasto 33 34.575 26.415
11 Violoncello arco Viola arco -7 -3.2 22.813
12 Click Hi hat 2 1.65 8.932
1.6,9,12 Synchronization of sounds with short onset time -1 -2.056 9.125
2,3,8,10 Synchronization of sounds with short and long onset times 31.5 34.375 29.515
1-12 Total values of synchronization tasks 3 10.573 26.191

Tab. 3: Overview on the synchronization values from different groups of musicians for all the twelve tasks and for task 11
(Violoncello arco - Viola arco).

Musicians Data task 1-12 Data task 11


Main instru^ Mean difference physical Mean difference physical
ment Number of participants
ts onset in synchronized Standard deviation onset in synchronized Standard deviation
position ms position ms
Singers 6 112.72 27.57 -14.833 30.142
Guitarists 4 -0.25 16.40 -12.75 20.775
Drummers 4 11.58 20.71 2.5 36.051
Strings 7 14.15 34.05 -1.143 12.575
Brass 8 11.03 25.69 1.875 23.314
Pianists 8 9.03 24.99 -3.375 21.902
others 3 13.89 21.26 7.333 16.921
Total 40 10.57 26.19 -3.2 22.813
102 103

statistically significant impact on the results; participants taking more time did not Short onset rise times
systematically differ from those taking less time (F=1.77, p=0.18). No influence The first and second assumptions are confirmed by the results of task 1, in which the
could be detected concerning the time of day when the experiment took place; there click (with an onset rise time of 0.07 ms) had to be synchronized with a ride stimulus
were no significant differences between the average responses of all tasks from mu- (with an onset rise time of 1.77 ms). The synchronizing of these two sounds with
sicians participating in the morning and those in the afternoon (F=0.95, p=0.33). very short onset times and different durations lead to a mean of 0.625 ms between
Also, there were no significant differences between Bachelor and Master program physical onsets. Likewise, the mean difference of physical onsets of-2.1 ms from all
students (F=0.09, p=0.77). Likewise, the number of years a participant had played tasks with sounds of short rise time supports our hypotheses. A fine-timed auditory
his main instrument had no significant influence on the results (F=0.16, p = 0.69). perception system enables humansunder special circumstancesto perceive asyn-
While female and male participants performed significantly different on the aver- chronies of at least 2 ms between the physical onsets of musical tones. The present
age of all tasks (F=6.21,p<0.05), the gender difference accounted for only 0.7% results are in accordance with previous findings (see Reuter 1995:36).
of the explained variance and therefore can be considered to have had only a very
weak effect. In sharp contrast, the varying test results between the different stimuli Masking
combinations of the twelve tasks accounted for at least 45% of the explained vari- Discussion is required for the negative mean values of physical onset differences
ance in all models. Hence, the specific synchronization task at hand is the single most in task 5 to 7 (table 2). In these tasks, participants placed a sound with short onset
important factor explaining the observed differences between the physical onsets of rise time before a sound with a long onset rise time. The synchronizing of a ride to
the synchronized sounds. a contrabass pizzicato sound, for example, resulted in a mean difference of physical
Variations in results between groups of musicians playing the same instrument onset of 3.9 ms, the ride sound's PMO placed slightly before the physical onset of the
(respectively the same class of instruments) were explored (table 3). Although the contrabass. According to our third hypothesis, the contrabass' PMO lies before it is
small and unequal number of participants in each group precludes the drawing of even acoustically present. These odd results are best explained by the possibility that
any firm conclusions, results point to possible significant differences of expertise some results of the synchronization tasks were influenced by perceptual limitations
between the instrumentalists (F = 3.48, p < 0.01). The group of guitarists, whose re- in the judgment of temporal order (Pastore and Farrington 1996), as participants oc-
sults differed significantly from those attained by other groups (t=-3.33, p<0.01), casionally commented on difficulties in deciding which of two synchronized sounds
shows the smallest standard deviation averaged over all tasks (16.4 ms), whereas the they heard first. The main influence for the negative values, however, is likely to
string group shows the highest standard deviation (34.05 ms). Interestingly, this high arise from temporal masking effects. Masking has also been reported by Gordon
standard deviation of the strings does not apply to task 11, in which bowed string (1987:93f.) in his experiment, where masking caused participants not to synchronize
tones had to be synchronized. The seven string players synchronized the physical sounds, but rather to "fuse" them. In the concerned tasks 5 and 7 of our experiment
onsets of these two similar sounds with the lowest standard deviation and with a this fusing may be explained as the strategy of participants to avoid masking by the
difference value of physical onsets of 2.1 ms, which was the second closestafter contrabass sound with its high intensity (see Hove et al. 2007:706). By positioning
the pianiststo the mean difference of onsets for all participants. click and ride sounds slightly before the contrabass sound, no masking effects af-
fected their audibility.

'Discussion Cross checking results


Related to our third hypothesis, a comparison between values obtained in synchro-
The current experiment was designed around a set of three assumptions. First, that nization tasks with two instrumental sounds and values from tasks with clicks al-
the onset rise time of the amplitude envelope is a main determining factor for a lows a cross checking of the results. If all results were generally valid, the sounds'
sound's PMO. Second, the PMO of a sound occurs before the amplitude envelope PMO should coincide in different combinations. For instance, the result from task
reaches its maximum level. Third, by synchronizing a click to an instrumental 11 (Violoncello arco as movable sound, Viola arco as fixed sound) should ideally
sound, the time difference of physical onsets would determine the PMO of that equal the positive difference of values from task 8 (Click as movable sound, Vio-
instrumental sound. loncello arco as fixed sound) and task 3 (Click as movable sound, Viola arco as
fixed sound). By subtracting the higher from the lower value of tasks 8 and 3, and by
104 105

W IT, Ifl VI V) M-a vi (^ N (^ cO


o m01^QOift
comparing this difference with the value from task 11, we get an error divergence c n
Nc oN\ cO;
O i> r^j 0(
N ^; -^
3 oc oo so c-i o <N -ci
of less than 9 ms. 01 S
Table 4 gives the values for all combination tasks with instrumental sounds in P4 (N
cp-l -J'j
comparison with calculated values from click tasks. These calculated values (II) are v
oc vt viesinesincs
i r~- 3
a* E
within the range of standard deviation of the synchronization task with the two in- E
3 S 5= O i
strumental sounds (I). Divergences between the two values (III) are relatively small, 5U .y
5 o
ranging from 3.0 to 8.7 ms. ca u
5s IO Vl O Vl VI
"w >1
rr
r i r-
" _: <N -*
Distribution of results s i a.
.,,o
The musicians' responses follow a unimodal distribution in all the tasks, contrary to E 5u o
the data obtained by Gordon who reports bimodal distributions (Gordon 1987:92- O^O VI - ^ a
100). However, the present data demonstrates marked differences in skewhess, a u
e V
kurtosis, and dispersion (figure 1).
It is notable that in tasks with clicks as movable sounds, the greater the fixed 1 a ',t- N >0 t- V2
sound's onset rise time, the higher, the distribution of the results, measured as stand- 1 3 " T O O s F t^ io o\ sD W
ard deviation (table 5 and figure 2). This may not apply for the group of sounds o r~. r^ h a vi a N
USE a o -' in t-^ r-^ TT 6 vd
with short onsets, but holds for sounds with onset rise times above the piano sound iO ^ - N M VI OO ^o
S ^3
o0n

8
r- i vivi5r-m
OOOOO
VI ^OMTiN
eS = Tf D ono m
a\ M
r~ m
Mvo

a

.5 -P 1.1B -* o> HI r- r*>


g s 3 r-; ; VI vi oq
3> G
d> c*"i r
N^ ts
c -Nvovinooo
1 ^ X
<Uow
=1
"go fc-
S3 a
b ><;
o a.
_ ! O tc
s
~5 e rt ^
9 s -o H
> U>
Ci, ^ c
e

Fig. 1: Density distribution (histogram) and functions for each task related to the physical
onset differences in ms (abscissa) and a normalized probability density scale (ordinate).
to O o= 21,
Grey bars represent empirical datafromour test sample. The curve represents the kernel UaI> Q p aw a
vS - ra c _
Xi
density function which is a non-parametric way of estimating (extrapolating) the probabil-
ity density function for the entire population. cs Tt r- a\ oon
106 107

dressing the question of temporal perception. Participants of our experiment reported


difficulties in synchronizing a click and a Violin sul tasto sound with its continuously
increasing and meandering rise of the amplitude envelope, which is expressed in the
highest standard deviation of the participants' responses (figure 3). Differences in
onset characteristics also caused difficulties in the experiment conducted by Gordon,
where participants had to synchronize a percussion soundhaving an onset rise time
of less than 10 mswith instrumental sounds of 45 ms and more onset rise time
(Gordon 1987:94).
The difficulty in identifying a PMO in sounds with slowly rising amplitude en-
velopes may be related to past findings on the discrimination of rise time, being best
at the shortest rise times and decreasing with increasing rise time.6 Interpreting the
findings from a musical perspective, the standard deviation may be a value for the
acoustic ambiguousness in synchronizing musical sounds: the longer a sound's onset
rise time, the less precise the synchronization in acoustical terms. As the comments
2000 ms given by participants indicate, the synchronization of the click with a violin sul tasto
Fig. 2: Correlation between sounds' onset rise time (abscissa) and standard deviation of sound with its long onset rise time was not so much solved as a perception task, but
physical onsets by synchronizing a click to sounds (ordinate). rather, as an exercise of microtemporal aesthetic choice.

Amotu
ild* (dBI
OO
. O-i
Conclusion

The data, obtained in an experimental condition of focused listening, provides evi-


dence that the onset rise time of a musical sound is an important factor for musicians
in determining its PMO, and hence in synchronizing musical sounds. This can be
related to results from neuroscientific experiments in which differentiated cortical
responses have been observed for stimuli with differing onset rise times (Thompson
et al. 2009). Moreover, the present findings show that the longer a sound's onset rise
time, the less agreement on its PMO. In interpreting measurements of microtemporal
Fig.3: Visualization of the synchronization task 10 (Click- Violin sul tasto) by the Lucerne music structures with sounds of long onset rise times, this range of PMO music be
Audio Recording Analyzer LARA (www.hslu.ch/lara). The circles on the amplitude enve- taken into consideration.
lope (computed using the RMS of a non-overlapping window ofl ms) show the position-
ing of clicks by the forty participants in our experiment. Multiple values are indicated by
numbers, the mean value by thefilledcircle. The limits of the standard deviation are given
by vertical lines. Acknowledgements

Our research was supported by research funds from Lucerne University of Applied
of 27 ms. The overall bivariate correlation between sounds' onset rise time and Sciences and Arts (08-478). We would like to thank the music students at Lucerne
standard deviation of physical onsets amounts to r = 0.92 and is highly significant School of Music for their participation in this research project. A special thank you
(p< 0.001). goes to Claudia Emmenegger for her thoughtful suggestions on various stages of
Slowly rising sounds and especially the extreme appearance as "smooth rhythms" our research.
with blurred onset and offsets (Gjerdingen 1993) need special attention when ad-
108 109

References Harsin Charles Andrew (1997). Perceptual-center modeling is affected by including acoustic
rate-of-change modulations. Perception and psychophysics, 59/2, pp. 243-251.
Alen Olavo (1995). Rhythm as duration of sounds in Tumba Francesa. Ethnomusicology, Hove Micheal J., Keller Peter E., and Krumhansl Carol L. (2007). Sensorimotor synchroniza-
39/i, pp. 55-71. tion with chords containing tone-onset asynchornies. Perception and psychophysics, 69/5,
Ashley Richard D. (2002). Do[n't] change a hair for me: The art of jazz rubato. Music percep- pp.699-708.
tion, 19/3, pp.311-332. Howell Peter (1988). Prediction of P-center location from the distribution of energy in'the
Auhagen Wolfgang (2008). Rhythmus und Timing. In Herbert Bruhn, Andreas C. Lehmann, amplitude envelope. Perception and psychophysics, 43, pp.90-99.
and Reinhard Kopiez (eds.), Musikpsychologie: Das neue Handbuch. Reinbek: Rowohlt, Keil Charles (1995). The theory of participatory discrepancies: a progress report. Ethnomu-
pp.437-457. sicology, 39/1, pp. 1-19:
Benadon Fernando (2006). Slicing the beat: Jazz eighth-notes as expressive microrhythm. Kopiez Reinhard (2005). Experimentelle Interpretationsforschung. In Helga De la Motte-
Ethnomusicology, 50/1, pp.73-98. Haber and GUnther Rotter (eds.), Musikpsychologie (Handbuch der Systematischen Mu-
Benadon Fernando (2009). Time warps in early jazz. Music theory spectrum, 31/1, pp. 1-25. sikwissenschaft 3). Laaber: Laaber, pp.459-514.
Bjurling Johan (2007). Timing in piano music: A model of melody lead (Master's Thesis Langner Jbrg and Goebl Werner (2003). Visualizing expressive performance in tempo-
in Music Acoustics at the School of Media Technology). Stockholm: Royal Institute of loudness space. Computer music journal, 27/4, pp. 69-83.
Technology. Marcus Stephen Michael (1981). Acoustic determinants of perceptual center (P-center) loca-
Bregman Albert S. (1993). Auditory scene analysis: Hearing in complex environments. In tion. Perception and psychophysics, 30/3, pp. 247-256.
Stephen McAdams, and Emmanuel Bigand (eds.), Thinking in sound: The cognitive psy- Morton John, Marcus Steve, and Frankish Clive R. (1976). Perceptual Centers (P-centers),
chology of human audition. Oxford: Oxford University Press, pp. 10-36. Psychological review, 83/5, pp.405-408.
Collier Geoffrey L. and Collier James L. (2002). A study of timing in two Louis Armstrong Palmer, Caroline (1989). Mapping musical thought to musical performance. Journal of ex-
solos. Music perception, 19/3, pp.463-483. perimental psychology, human perception and performance, 15/12, pp. 331-346.
Cooper Anre" M., Whalen, D. H., and Fowler Carol A. (1986). P-center are unaffected by Pastore Richard E. and Farrington Shannon M. (1996). Measuring the difference limen for
phonetic categorization. Perception andpsychophysics, 39/3, pp. 187-196. identification of order of onset for complex auditory stimuli. Perception and psychophy-
Fowler Carol A. (1979). "Perceptual centers" in speech production and perception. Perception sics, 58/4, pp.5X0-526.
and psychophysics, 25/5, pp.375-388. Pfleiderer Martin (2006). Rhythmus: Psychologische, theoretische und stilanalytische Aspekte
Fox Robert Allen and Lehiste Use (1987). Effect of unstressed affixes on stress-beat location popularer Musik, Bielefeld: Transcript.
in speech production and perception. Perceptual and motor skills, 65, pp. 35-44 Plaisier Marco (2007). The perceptual moment of occurrence of simple sounds with an am-
Friberg Anders and Sundstrom Andreas (2002). Swing ratios and ensemble timing in jazz per- biguous onset (Master thesis, Universiteit Eindhoven, Department of Technology Manage-
formance: Evidence for a common rhythmic pattern. Music perception, 19/3, pp. 333-349. ment). Eindhoven: Universiteit Eindhoven, 2007.
Gabrielsson Alf (1999). The performance of music. In Diana Deutsch (ed.), The psychology Pompino-Marchall Bernd (1989). On the psychoacoustic nature of the P-center phenomenon.
of music. San Diego: Academic Press, pp. 501-602. Journal ofphonetics, 17, pp. 175-192.
Gerischer Christiane (2003). O suingue baiano: MikrorhythmischePhanomerie in baidhischer Progler Josef A. (1995). Searching for swing: Participatory discrepancies in the jazz rhythm
Perkussion (Europaische Hochschulschriften 36, Musikwissenschaft 237). Frankfurt a. section. Ethnomusicology, 3911, pp.21-54.
M.: Lang. Rapp-Holmgren, K. (1971). A study of syllable timing. STL-QPSR, 12/1, pp. 14-19.
Gjerdingen Robert O. (1993). "Smooth" rhythms as probes of entrainment. Music perception, Rasch Rudolf A. (1978). The perception of simultaneous notes such as in polyphonic musk.
10/4, pp.503-508. Acustica, 40, pp. 21-33.
Goebl Werner (2001). Melody lead in piano performance: Expressive device or artifact? Rasch Rudolf A. (1979). Synchronization in performed ensemble music. Acustica, 43,
Journal of the Acoustical Society of America, 110, pp. 563-572. pp.121-131.
Goebl Werner and Palmer Caroline (2009). Synchronization of timing and motion among Repp Bruno H. (1996). Patterns of note onset asynchronies in expressive piano performance.
performing musicians. Music perception, 2615, pp.427-438. Journal of the Acoustical Society of America, 100/6, pp. 3917-3932.
Goebl Werner and Parncutt Richard (2002). The influence of relative intensity on the per- Repp Bruno H. (1997). Expressive Zeitgestaltung in einem Prelude von Debussy: Ein Ver-
ception of onset asynchronies. In Catherine Stevens, J. Renwick, Gary E. McPherson, D. gleich zwischen Klavierschulern und Experten. Musicae scientiae, 1/2, pp.257'-268.
Burnham, and Emery Schubert (eds.), Proceedings of the 7h International Conference on Reuter Christoph (1995). Der Einschwingvorgang nichtperkussiver Musikinstrumente: Aus-
Music Perception and Cognition (Sydney, Australia, 2002). Adelaide: Causal Productions, wertungphysikalischer undpsychoakustischer Messungen. Frankfurt a. M.: Lang.
pp.613-616. Rose, Richard F. (1989). An analysis of timing in jazz rhythm section performance. Austin:
Gordon John W. (1987). The perceptual attack time of musical tones. Journal of the Acousti- University of Texas Press.
cal Society of America, 82, pp.88-105.
>&

110

Rosen Stuart Michael and Howell Peter (1981). Plucks and bows are not categorically per-
ceived. Perception and psychophysics, 30/2, pp. 156-168.
Schiitte Hermann (1978). Ein Funktionsschema fur die Wahrnehmung eines gleichmassigen
Rhythmus in Schallimpulsfolgen. Biological Cybernetics, 29, pp.49-55.
Scott Sophie K. (1998). The point of P-centres. Psychological research, 67/i,pp.4-ll.
Senn Olivier, Kilchenmann Lorenz, and Camp Marc-Antoine (2009). Expressive timing:
Martha Argerich plays Chopin's Prelude op.28/4 in E minor. In Aaron Williamoh, Shar-
man Pretty, and Ralph Buck (eds.), Proceedings of the International Symposium on Per1
formance Science 2009. Utrecht: European Association of Conservatoires, pp. 107-112.
Shaffer L. H. (1984). Timing'in solo and duet piano performances.- Quarterly journal of ex-
perimental psychology, 36/4, pp. 577-595.
Sundberg Johan and Bauer-Huppmann Julia (2007). When does a sung tone start? Journal of
voice, 21/3, pp.285-293.
Terhardt Ernst and Schiitte Hermann (1976). Akustische Rhythmus-Wahrnehmung: Subjek-
tive Gleichmassigkeit. Acustica, 35, pp. 122-126.
Thompson Jennifer M., Goswami Usha, and BaldewegTorsten (2009). The ERP signature of
sound rise time changes. Brain research, 1254, pp.74-83.
van Heuven V.J.J.P. and van den Broecke M.P.R. (1979). Auditory discrimination of rise
and decay times in tone and noise bursts. Journal of the Acoustical Society of America,
66/5, pp. 1308-1315.
Vos Joos and Rasch Rudolf A. (1981). The perceptual onset of musical tones. Perception and
psychophysics, 29/4, pp.323-335.
Vos Joos, Mates Jiri, and van Kruysbergen Noud W. (1995). The perceptual centre of a stimu-
lus as the cue for synchronization to a metronome: Evidence from asynchronies. Quarterly
journal of experimental psychology, 48/4, pp. 1024rl040.
Waadeland Carl Haakon (2001). It don't mean a thing if it ain't got that swing: Simulating ex-
pressive timing by modulated movements. Journal of new music research, 30/1, pp. 23-37.

Notes
1 see Auhagen 2008; Gabrielsson 1999; Keil 1995; Pfleiderer 2006; Repp 1997 for overviews.
2 Bjurling 2007; Friberg and Sundstrom 2002; Gerischer 2003; Goebl 2001; Goebl and Palmer 2009;
Goebl and Parncutt 2002; Palmer 1989; Rasch 1978,1979; Repp 1996; Senn et al. 2009; Shaffer
19841.
3 Fowler 1979; Fox and Lehiste 1987; Harsin 1997; Howell 1988; Marcus 1981; Pompino-Marschall
1989; Rapp-Holmgren 1971; Scott 1998.
4 Hove et al. 2007; Plaisier 2007; Schiitte 1978; Terhardt and Schiitte 1976; Vos and Rasch 1981;
Vosetal. 1995:1026f.
5 for example Men 1995: 57; Benadon 2006:76,2009:5; Progler 1995:32-46.
6 Gordon 1987:88-92; Reuter 1995:33f.; Rosen and Howell 1981; van Heuven and van den Broecke
1979.

You might also like