Johan Sundberg. Intonation in Singing.

Intonation in Singing
Oxford Handbooks Online

Johan Sundberg
The Oxford Handbook of Singing
Edited by Graham F. Welch, David M. Howard, and John Nix
Subject: Music, Applied Music Online Publication Date: Jan 2016

DOI: 10.1093/oxfordhb/9780199660773.013.36
Abstract and Keywords
Intonation is an important aspect of sung performances. This chapter presents some

investigations of vocal ensemble and solo singing. The magnitude of the standard
deviation of F0 (SDF0) in a choral section of a high-quality choir is discussed, as are
intonation effects of common partials, the presence of high-frequency partials, and
vibrato-free tone production, which all tend to reduce SDF0, i.e., to facilitate intonation.
Other topics include accuracy, expressivity, and the role of equally tempered tuning of
accompanying instruments. Reaction time for F0 correction of intonation errors is slow,
suggesting that copying the pitch from fellow singers in an ensemble is not a useful
strategy. The F0 pattern used in legato performances of sequences of short notes is
described as well as solo singers’ intonation strategies in different musical and emotional
contexts. Likewise, professional music listeners’ tolerance to intonation inaccuracy seems
to vary depending on context.
Keywords: vibrato, solo singing, ensemble singing, common partials, accuracy, expressivity, reaction time, equally
tempered tuning
Background
In music composition as well as in music performance the musical scale plays a
fundamental role. From a physical point of view, a scale corresponds to a division of the
frequency continuum into discrete steps, the scale tones. In our Western music culture,
there are mostly seven of them in the octave, and the frequencies of the scale tones in
other octaves are obtained by frequency doubling and halving.
In the development of these scales the widespread use of simultaneous playing of several
instruments that produce harmonic spectra seems to have played an important role.
Under these conditions some of the lower partials of two simultaneously sounding tones
Page 1 of 15
PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). © Oxford University Press, 2018. All Rights
Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in
Oxford Handbooks Online for personal use (for details see Privacy Policy and Legal Notice).
Subscriber: Mount Sinai School of Medicine; date: 03 February 2019

will coincide. This happens when their frequencies can be expressed as a ratio between
small integers.
For example, if an instrument producing harmonic spectra plays a tone with a

fundamental frequency (F0) of 110 Hz, its overtones appear at 220, 330, 440, 550, 660,
770, 880, 990, etc. Hz. If another instrument simultaneously plays a tone a pure fifth
higher, i.e., with a F0 of 1.5*110 = 165 Hz tone, the overtones of this tone will appear at
330, 495, 660, 825, 990, etc. Hz. In other words, if the frequency ratio is 165:110 = 3:2,
which corresponds to a pure fifth, every third partial of the lower tone coincides with
every second partial of the upper tone.
Coinciding partials results in a special timbral quality, referred to as consonance, and

absence of coinciding partials induces the opposite timbral quality, dissonance. Thus, the
interval of a fifth sounds consonant. The principle is that intervals with frequency ratios
that can be written as ratios between low integers, i.e., 2:3, 3:4, 4:5, etc., sound
consonant, and the lower the integers, the greater the number of coinciding harmonics,
and the more consonant the interval. The octave therefore is the most consonant interval,
as all partials of the upper tone are already present in the spectrum of the lower tone. A
condition for this to happen is that the intervals are played exactly in tune. If not, they
generate beats.
Dissonance is the other extreme. Maximal dissonance occurs when two partials are
separated by an interval that equals a quarter of a critical band of hearing. These bands
represent a type of analysis bandwidth of the ear (Plomp and Levelt 1965). For
frequencies above about 500 Hz the critical bands are about a minor third wide. For
lower frequencies it is around 100 Hz.
In this sense, our Western diatonic scale offers possibilities to play both very consonant
and very dissonant intervals. It is tempting to speculate how music would have sounded if
it were always played on instruments that generated inharmonic partials.
Consonance of intervals with coinciding partials happens when the intervals are tuned so
that their frequency ratios can be written as ratios between small integers. This tuning is
generally referred to as either just or pure. If the same intervals are played with almost,
but not exactly the same, frequency ratios, beats appear. The reason is that the overtones
do not coincide exactly. Thus, just tuning is the only one that does not give rise to beats in
consonant intervals. On the other hand, in playing solo music, just tuning is often not
applied. The greatest difference happens on the intervals that exist in both major and
minor versions: the second, third, sixth, and seventh. Rather, the major versions are
widened and the minor versions are narrowed. This tuning is referred to as the
Pythagorean.
The frequencies of the scale tones in Pythagorean tuning can be obtained by piling pure
fifth intervals on top of each other and then reducing the frequencies by halving, such
Page 2 of 15

that they all arrive in the same octave. This tuning enhances the difference between
major and minor versions of intervals.
The equally tempered tuning (ETT) can be regarded as a brutal mathematical method for
obtaining the frequencies of the scale tones that lie between those of the pure and the
Pythagorean tunings. In the ETT, the octave interval is simply chopped into twelve
intervals, all of exactly identical width. Here, all consonant intervals except the octave
have only nearly, but not exactly, coinciding partials. Hence they generate beats when the
tones sound simultaneously.
Figure 1 compares pure

and Pythagorean tunings
with the equally tempered
tuning. Pure tuning
reduces the contrast
between intervals that
exist in both minor and
major versions, in the
Figure 1 Deviations, in cent, from the equally sense that the minor
tempered tuning in the pure and in the Pythagorean
tunings.
versions of the second,
third, sixth, and seventh
are played slightly wide, and the major versions slightly narrow. The Pythagorean has the
opposite effect, enhancing the contrast between minor and major intervals.
Intonation in ensemble singing

Choir
Obviously, the accuracy of intonation is limited in all kinds of music performance,

including choral ensembles. The accuracy can be measured in terms of the standard
deviation of the F0 values obtained from members of an ensemble.
Page 3 of 15

Ternström and Sundberg

(1988) glued
accelerometers, which are
microphones that are
sensitive only to
vibrations, to the throats
of six bass singers in a
choir of good reputation
and asked them to perform
a simple eight tones
cadence four times after
first having rehearsed it.
Figure 2 Standard deviations measured in the six The maximum differences
bass singers of a good amateur choir singing the between the six subjects’
bass part of the cadence shown four times.
mean F0 (MF0) varied for
different tokens between 5
and 66 cents for the various tones. The overall distribution of deviations from the group
means for the eight tones was close to normal. As shown in Figure 2, most values of the
standard deviations for the different tones lay within an interval of ±15 ± from the group
mean. The average standard deviations of the eight tones were 13 cents, with extremes at
10.3 and 15.8 cents. Thus, two-thirds of the F0 values produced by these basses agreed to
within ±13 cents (corresponding to about 1 Hz at 140 Hz F0). The largest deviation from
the group average produced by individual singers was on the order of +45 cents. It is
possible that smaller scatter values would have been obtained in the other choir sections,
as they sing in a higher frequency range where similar errors will cause quicker and thus
probably more apparent beats.
It is tempting to assume that just intonation is preferred in ensemble singing such that
beats are avoided. On the other hand, beats may not be a really severe disadvantage,
since they are produced only when the F0 is perfectly constant, a condition only met in
electronically generated tones. Further, as mentioned, just tuning has the disadvantage of
being associated with intervals between successive tones that are far from the
Pythagorean tuning, which seems preferred in performances of solo parts and melodies.
In the same study, Ternström and Sundberg studied also the relevance to intonation of
three factors: (1) vibrato, (2) amplitude of common partials, and (3) amplitudes of partials
above the first common partial. They presented synthesized tones with all combinations
of these three properties to each of eighteen male choir singers who were asked to sing a
major third and a fifth above the reference tones, which were presented over a
loudspeaker. The standard deviation of the F0 of the tones produced was measured.
Page 4 of 15

The results are presented

in Figure 3, where
significant effects are
marked with asterisks and
the different combinations
with lower-case letters. As
can be seen in the figure,
all these three tone
properties had significant
Figure 3 Standard deviations, in cent, of F0 values effects on how accurately
observed when eighteen male choral singers sang a the major third and the
fifth and a major third above synthesized reference
tones with and without the indicated properties. fifth were sung, but the
Effects significant at the 5 percent, 2.5 percent, and effect of a given property
1 percent levels are marked with one, two, and three
asterisks, respectively.
depended on what other
properties the tone
From Ternström and Sundberg (1988).
possessed. In any event,
the results showed that the
presence of common partials, an absence of vibrato, and the presence of high partials
may facilitate intonation under certain conditions. In addition, it was found that
intonation was facilitated when the lowest common partial was enhanced by a formant.
Jers and Ternström (2005) analyzed intonation in a choir of sixteen singers. The ensemble
performed in unison a music example containing half notes and quarter notes, and the
males sang one octave below the females. The melody was performed in a slow and a fast
tempo, average quarter note durations being about 500 ms and 750 ms. Their F0 was
measured from accelerometer microphones glued to the singers’ noses and recorded on
separate tracks of a tape recorder. F0 was measured as the average across the entire
tone. The result showed that the standard deviation of F0 of the ensemble decreased from
around 25 cents to 16 cents in the slower tempo.
This effect can be assumed

to reflect the time it takes
for a singer to adapt to an
external F0 reference.
Grell and associates
measured this reaction
time in members of a
highly skilled and a
moderately skilled choir
Figure 4 Distribution of reaction times for highly (Grell et al. 2009). They
and moderately skilled singers (upper and lower were presented with two
panels).
simultaneous synthesized
After Grell et al. (2009).
sung vowels forming a fifth
interval. The tones were
Page 5 of 15

presented over loudspeakers. Suddenly, the tones were shifted by 50 or 100 cents upward
or downward and the task was to adapt to this shift in reference.
Great inter-individual variation was observed, but the most common reaction time was
close to 150 ms, as can be seen in Figure 4. Thus, it took about 150 ms for a singer to
adjust tuning to a heard reference, for example the tones from the fellow singers in a
choir. This seems to explain the findings regarding the effect of tempo in the Jers and
Ternström study. In that experiment the fine-tuning of the tones was averaged across the
entire tone, so the mean included the fine correction of tuning that may have happened
during the first 150 ms.
Barbershop
Given these results barbershop singing is an interesting case. It is traditionally performed

by four singers, of which one, the lead singer, uses falsetto register, while the other
singers use modal register. All singers strive to sing without vibrato, so intonation could
be expected to be close to just. This hypothesis was tested in a study where two well-
renowned barbershop groups were recorded by means of accelerometers glued to the
external surface of the neck near the trachea (Hagerman and Sundberg, 1980). They
sang examples that belonged to their standard repertoire. The results showed that the
lead voice served as the tuning reference for the group. The standard deviations of F0
ranged between 4 and 17 cents, thus much less than that found in choral singing. The
intervals between simultaneous tones were found to deviate systematically from their
corresponding values in both just and Pythagorean intonation. The major third in major
triads with the function of a dominant was found to be a slightly stretched version of a
pure third, while in a tonic chord it was closer to a flattened version of a Pythagorean
third. The major third in a minor triad showed the same width as the major third of major
triads which included the seventh. Interestingly, the deviations from just intonation did
not give rise to beats, presumably due to the finite degree of periodicity of the tones.
Vibrato
Singing in the classical Western operatic tradition includes vibrato as an important
property. It corresponds to a quasi-sinusoidal modulation of F0. As a consequence, the
frequencies of the harmonic partials also vary in phase with F0. This also implies that the
partials which undulate in frequency prevent the production of beats in mistuned, i.e.,
non-pure intervals. This makes intonation in singing in the classical Western operatic
tradition a particularly interesting area.
A question of basic relevance is what pitch is perceived when a tone has such an
undulating F0. Experiments where musically trained subjects have matched the pitch of a
synthetic sung vibrato tone with the same tone void of vibrato have indicated that the
pitch corresponds to the mean F0, averaged over a complete vibrato period (Horan and
Shonle 1980; Sundberg 1981). This is true for vibrato rates in the range of about 5 to 7
Hz, which is typical in professional Western opera singing. Thus, the perceptual system
Page 6 of 15

appears to calculate a running average of F0 over a time window approximately fitting

one vibrato period, that is, about 150 to 200 ms.
Figure 5 Fundamental frequency pattern produced

by a professional singer performing the sequence of
sixteenth notes shown.
A striking application of
this averaging is provided
by the F0 pattern observed
in legato performances of
coloratura passages, i.e.,
sequences of short tones
which are not separated by
pauses. Figure 5 shows a
typical example. Each
scale tone corresponds to
an F0 pattern encircling
Figure 6 Green curve: fundamental frequency the target F0. A similar
pattern resulting from superimposing a sinusoidal pattern can be obtained if
vibrato on a glissando represented by the black
dashed curve. The black curve shows a running a vibrato-like sinusoidal
average of the green curve. modulation is
superimposed on a
glissando, as illustrated in Figure 6. If this pattern is processed with a running average
function with a window of about 200 ms width, a quasi-stepwise changing F0 is obtained,
as illustrated in the same figure.
Page 7 of 15

In fast staccato singing,

the voiced segments of the
tones are sometimes
shorter than a complete
vibrato cycle. Figure 7
shows an example. The
pitch perceived of such
short tones that contain
incomplete vibrato cycles
has been studied in detail
by d’Alessandro and
Figure 7 Fundamental frequency pattern produced
Castellengo (1994). They
in staccato performance of an excerpt from a
commercial recording of the aria Der Hölle rache used synthesized vibrato
from W.A. Mozart’s opera Die Zauberflöte. The target tones where the vibrato
values are marked with heavy green lines.
cycle was interrupted at
different places. The
results showed that the pitch perceived corresponds to an F0 average where more weight
was placed on the final part of the event. Thus, the overall F0 pattern influences pitch
perception. The authors also showed that the F0 of the perceived pitch can be
approximated by an equation.
Provided that the rate lies within about 5 and 7 Hz, vibrato generally has no influence on
the accuracy of the pitch perceived. Thus, in an experiment, musically educated listeners
adjusted the F0 of a vibrato free synthetic vowel such that it appeared to have the same
pitch as the same vowel with vibrato. The accuracy was found to be the same as when the
same subjects repeated the experiment with vowels that lacked vibrato. On the other
hand, van Besouw and associates (2008) reported that the tolerance of what was
considered to be in tune was somewhat more generous for vibrato tones than for vibrato
free tones. Absence of beats between vibrato tones may contribute to this effect.
Accuracy
Listeners’ accuracy of pitch perception is obviously crucial to the accuracy required in
singing. This accuracy has been analyzed by Vurma and Ross (2006). They analyzed
professional singers’ performances of ascending and descending versions of three
intervals: minor second, triton, and fifth. The results showed that, on average, the
singers’ intonation was very close to the equally tempered tuning; the F0 averages had a
standard deviation in the vicinity of 20 cents. Thus, the singers’ mean F0 values were less
than 20 cents from equally tempered tuning. Yet there was a systematic tendency to
expand the wider intervals fifth and tritone and to compress the narrow minor second
interval slightly.
While Vurma and Ross investigated singers’ intonation of isolated intervals, deviations
from ETT seem also to depend on the musical context. Sundberg (2011) analyzed the
Page 8 of 15

intonation in commercial recordings of Radames’ recitative preceding the “Celeste Aida”

in the first act of Verdi’s opera Aida, as performed by tenors Carlo Bergonzi, Jussi
Björling, Placido Domingo, Luciano Pavarotti, and Richard Tucker.
Most of these singers

tended to sing the higher
tones sharper than the
lower tones, a tendency
also observed in
performances of
instrumental music
(Friberg 1991). For the
Figure 8 The indicated singers’ deviations from their phrase-peak tone A4 in bar
F0 average of the sustained tone in bar 21. Each 21, a general tendency
data point corresponds to the mean of half a vibrato
was observed to sing
cycle measured at its peak and valley.
somewhat sharp relative to
From Sundberg (2011).
the ETT of the
accompanying orchestra.
As illustrated in Figure 8, some of the singers started the tone somewhat flat and
gradually sharpened it towards the end of the tone, so that the following descending
octave leap was stretched. The sharpening reached between 10 and 20 cents at the end
of the tone. As compared with the ETT of the accompanying orchestra, Björling’s
intonation of the phrase-peak tone A4 in bar 21 was no less that 60 cents sharp in a live
recording.
Given these observations, a relevant question is the relationship between a singer’s

intonation and music listeners’ perception of a singer being out of tune. Sundberg and
associates (1996) studied the accuracy of intonation in a musical context, performances of
Franz Schubert’s song Ave Maria. F0 data was collected from ten commercial recordings
in terms of the F0 values of adjacents maxima and minima in a sequence of complete
vibrato cycles. Then the averages were calculated across each of the tones longer than an
eighth note in the song. In calculating these ETT values, the tuning of the accompaniment
was used as the reference. Thus, if the tuning of the pitch of A4 was 443 Hz, the ETT
values for the tones were calculated using 443 Hz as the reference. The results showed a
great variation, ranging from 42 cents sharp to 42 cents flat. Thus, much greater
deviations from ETT occurred in a musical context than when professional singers
performed isolated intervals, as in the Vurma and Ross study (2006).
Page 9 of 15

A relevant question is to
what extent these
deviations were consistent.
Nine of the singers had
recorded both the first and
the second verse of the
song, thus allowing a
comparison of intonation
of the same tone. The
comparison was carried
out for tones 1, 2, 3, 23,
24, and 25, i.e., the first
and last three long tones
of the song. The result is
shown in Figure 9, where
Figure 9 Examples of the consistency of intonation
the dashed lines represent
observed in two singers’ performance of the three a ±10 cents difference. In
long tones of the first phrase in verse 1 and verse 2 forty-eight tones out of the
(dotted and solid curves) of Franz Schubert’s Ave
Maria. total of (6 x 9) fifty-four
From Sundberg et al. (1996).
tones, the difference
exceeded ±15 cents. One
singer showed a clear tendency to sing the tones in the second verse flatter than those in
the first verse (open triangles in the graph). Thus, in most cases the deviations from the
ETT were reasonably consistent.
A listening test was carried out with six highly experienced professional music listeners,
representing different professions: teacher of singing, phonogram producer, choral
conductor, and piano accompanist. They were provided with one score for each of the ten
singers and were asked to circle all tones, which they perceived as out of tune.
Page 10 of 15

The result showed a

surprisingly great scatter.
Many notes were marked
as out of tune by three or
four experts but not by the
others. Figure 10 shows
examples of intonation
that were classified as out
of tune by none of the
experts and by at least five
experts. Of the former
tones most, but not all,
were quite close to ETT,
but some were clearly
Figure 10 Examples of intonation curves observed in
commercial recordings of Franz Schubert’s Ave sharper. Of the tones
Maria. The numbers refer to the averaged deviation perceived as out of tune,
in Cent from ETT. Six expert listeners perceived the
examples in the left column as in tune and the many showed an F0 curve
examples in the right column as out of tune. below the ETT value.
After Sundberg et al. (1996).
The mean F0 values for

the 25 long tones of the
song are shown in Figure
11. Triangles refer to tones
classified as out of tune by
at least five of the expert
listeners, and tones that
none of the experts
complained about are
represented by diamonds.
Figure 11 Mean F0 of the long tones of Franz In most cases, the
Schubert’s song Ave Maria, measured in commercial intonation values
recordings. The horizontal dashed line represents
the ETT as related to the tuning of the
perceived as in tune varied
within a band of 20 cents.
accompaniment. Green diamonds and red triangles
represent cases perceived as in tune and out of tune
However, the center of this
by a panel of expert listeners.
band deviated from the
From Sundberg et al. (1996).
ETT value in some cases,
for example for tone
number 3. None of the expert listeners complained about the flat intonation of the last
tone of the piece.
Notes number 6, 7, 18, and 21 showed exceptionally great variation of intonation of tones
that none of the expert listeners perceived as out of tune, varying no less than between
35 cents sharp and 20 cents flat for tone 7. The reason for this is an open question, but it
might be relevant that this tone is a suspension note, presenting the fourth that moves to
Page 11 of 15

the major third in the next beat. Moreover, the note initiates a chromatically falling
sequence of the following melody notes that appear on the stressed position of the beats.
Intonation and expression

These observations suggest that intonation may serve expressive purposes. This was
studied in an investigation by Sundberg and associates (2013). The material was a set of
recordings where an internationally prominent baritone, Håkan Hagegård, sang eleven
music excerpts from the Lieder repertoire in two ways, as in a concert and as void of
musical expression as he could, the neutral version. Six excerpts had an excited character
and five a peaceful, contemplative character. Intonation was compared with the ETT
related to the mean intonation of the entire excerpt.
A marked intonation
difference was revealed
between the excited and
the peaceful excerpts. In
the concert versions of the
excited examples, the
phrase-peak tones were
sung about 25 cents sharp,
on average. In the peaceful
examples the sharpening
was close to zero in both
versions. Moreover, in the
excited examples, the
sharpening was greater in
the concert versions than
Figure 12 Deviations from ETT for the phrase-peak
in the neutral versions, as
tones in the singer’s neutral and concert versions of
the excerpts. Triangles and circles refer to excited illustrated in Figure 12.
and peaceful excerpts, respectively. This supported the
From Sundberg et al. (2013). assumption that the
sharpening was used for
expressive purposes in the excited excerpts.
To test this assumption the sharpening of the phrase-peak tones in the excited examples
was eliminated using the Melodyne software. Thus, the intonation of these tones was
flattened such that the mean F0 became in accordance with ETT. The original version and
the manipulated version of the excited examples were then presented pair-wise to
musicians, who were asked to decide which version in the pair sounded more expressive.
The result showed that the original versions, having sharpened phrase-peak tones, were
perceived as significantly more expressive than the manipulated.
Page 12 of 15

The results of this experiment was in accordance with the assumption that tuning can be
used as an expressive tool and that sharpening phrase-peak tones can be used for this
purpose in excited contexts. It can also be noted that the intonation of an interval may
carry some emotional information.
In view of these investigations, a relevant question is how small the intonation differences
are that music listeners can notice. The answer appears to depend on the musical context
and on the listening conditions.
The just noticeable difference in pitch between two tones presented in sequence
approaches 10 cents. In a listening experiment Friberg and Sundberg (1994) presented
three versions of short musical excerpts to musically experienced students. In one of the
versions the tuning of the scale tones deviated from ETT, and the subjects’ task was to tell
which of the three presentations was different from the other two. When an incorrect
answer was obtained the deviations were made twice as large in the subsequent
presentation. The results suggested that the just noticeable difference in tuning for these
listeners was about 42 cents, i.e., about four times larger than what has been found in
comparison of single tones presented in succession. Thus, it seems that even musically
experienced listeners may completely lack sensitivity to fine-tuning, their pitch
discrimination being limited to discriminate nothing but semitones. By contrast,
professional music listeners, such as teachers of singing and phonogram producers, seem
to have a tolerance zone of ±10 cents, as suggested by the results shown in Figure 10.
Thus, experience and education appears to sharpen the auditory system with regard to
fine-tuning as well. An improved ability to detect tuning differences may also increase a
listener’s possibilities to perceive expressive components of a musical performance.
The study of phrase-peak tone intonation in sung performance of excited music excerpts
suggested that sharpening such tones increased the expressiveness (Sundberg et al.
2013). Some experimental findings indeed support the hypothesis that intonation may
add an emotional coloring even to an octave interval (Makeig and Balzano 1982). In any
event, it seems fair to assume that an increased accuracy of pitch perception paves the
way for a more complete experience of emotional colors embedded in music
performances.
It is tempting to make a final remark regarding singing in tune and singing out of tune. In
today’s recording studios software packages are available that allow manipulation of
vocal artists’ pitch. If such corrections are made on the assumption that deviations from
ETT equals singing out of tune, they may in fact have the effect of reducing the
expressivity of the singer’s performance. It seems important to keep in mind, that ETT
cannot be accepted as the gold standard for tuning.
References
d’Alessandro, C. and Castellengo, M. (1994). The pitch of short-duration vibrato tones.
Journal of the Acoustical Society of America 95: 1617–1630.
Page 13 of 15

van Besouw, R.M., Brereton J.S., and Howard, D.M. (2008). Range of tuning for tones with
and without vibrato. Music Perception 26: 145–155.
Friberg, A. (1991). Generative rules for music performance: a formal description of a rule
system. Computer Music Journal 15(2): 56–71.
Friberg, A. and Sundberg, J. (1994). Just noticable difference in duration, pitch and sound
level in a musical context. In: Proceedings of the 3rd International Conference for Music
Perception and Cognition, pp. 339–340, Liège, July 23–27.
Grell, A., Sundberg, J., Ternström, S., Ptok, M., and Altenmüller, E. (2009). Rapid pitch
correction in choir singers. Journal of the Acoustical Society of America 126(1): 407–413
Hagerman, B. and Sundberg, J. (1980). F0 adjustment in Barbershop singing. Journal of

Research in Singing 4(1): 3–17.
Horan, J.I. and Shonle, K.E. (1980). The pitch of vibrato tones. Journal of the Acoustical
Society of America 67: 246–252.
Jers, H. and Ternström, S. (2005). Intonation analysis of a multi-channel choir recording.

TMH-QPSR 47(1): 001–006.
Makeig, S. and Balzano, G. (1982). Octave tuning—two modes of perception. Paper

presented at Research Symposium on Psychology and Acoustics of Music, University of
Kansas, Lawrence, April.
Plomp, R. and Levelt, A. (1965). Tonal consonance and critical bandwidth. Journal of the
Acoustical Society of America 38: 548ff.
Sundberg, J. (1981). Effects of the vibrato and the ‘singing formant’ on pitch. Journal of
Research in Singing 5(2): 5–17.
Sundberg, J., Prame, E. and Iwarsson, J. (1996). Replicability and accuracy of pitch
patterns in professional singers. In: P. Davis and N. Fletcher (eds), Vocal Fold Physiology:
Controlling Complexity and Chaos, pp. 291–306. San Diego: Singular.
Sundberg, J. (2011). Some observations on operatic singer’s intonation. Interdisciplinary

Studies in Musicology 10: 47–59.
Sundberg, J., Lã, F.M., and Himonides, E. (2013). Intonation and expressivity: A single
case study of classical western singing. Journal of Voice 27: 391e–397e.
Ternström, S. and Sundberg, J. (1988). Intonation precision of choir singers. Journal of the
Acoustical Society of America 84: 59–69.
Vurma, A. and Ross, J. (2006). Production and perception of musical intervals. Music
Perception 23: 331–344.
Page 14 of 15

Johan Sundberg
Johan Sundberg had a personal chair in Music Acoustics at the department from
1979 to his retirement 2001. Since 2002 he is Visiting Professor at the University of
London, UK.
Page 15 of 15

Johan Sundberg. Intonation in Singing.

Uploaded by

Copyright:

Available Formats

Johan Sundberg. Intonation in Singing.

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Johan Sundberg. Intonation in Singing.

Uploaded by

Copyright:

Available Formats

Intonation in Singing

Oxford Handbooks Online

Subject: Music, Applied Music Online Publication Date: Jan 2016

Abstract and Keywords

Intonation is an important aspect of sung performances. This chapter presents some

Subscriber: Mount Sinai School of Medicine; date: 03 February 2019

For example, if an instrument producing harmonic spectra plays a tone with a

Coinciding partials results in a special timbral quality, referred to as consonance, and

Subscriber: Mount Sinai School of Medicine; date: 03 February 2019

Figure 1 compares pure

Intonation in ensemble singing

Obviously, the accuracy of intonation is limited in all kinds of music performance,

Subscriber: Mount Sinai School of Medicine; date: 03 February 2019

Ternström and Sundberg

Subscriber: Mount Sinai School of Medicine; date: 03 February 2019

The results are presented

This effect can be assumed

Subscriber: Mount Sinai School of Medicine; date: 03 February 2019

Given these results barbershop singing is an interesting case. It is traditionally performed

Subscriber: Mount Sinai School of Medicine; date: 03 February 2019

appears to calculate a running average of F0 over a time window approximately fitting

Figure 5 Fundamental frequency pattern produced

Subscriber: Mount Sinai School of Medicine; date: 03 February 2019

In fast staccato singing,

Subscriber: Mount Sinai School of Medicine; date: 03 February 2019

intonation in commercial recordings of Radames’ recitative preceding the “Celeste Aida”

Most of these singers

Given these observations, a relevant question is the relationship between a singer’s

Subscriber: Mount Sinai School of Medicine; date: 03 February 2019

Subscriber: Mount Sinai School of Medicine; date: 03 February 2019

The result showed a

The mean F0 values for

Subscriber: Mount Sinai School of Medicine; date: 03 February 2019

Intonation and expression

Subscriber: Mount Sinai School of Medicine; date: 03 February 2019

Subscriber: Mount Sinai School of Medicine; date: 03 February 2019

Hagerman, B. and Sundberg, J. (1980). F0 adjustment in Barbershop singing. Journal of

Jers, H. and Ternström, S. (2005). Intonation analysis of a multi-channel choir recording.

Makeig, S. and Balzano, G. (1982). Octave tuning—two modes of perception. Paper

Sundberg, J. (2011). Some observations on operatic singer’s intonation. Interdisciplinary

Subscriber: Mount Sinai School of Medicine; date: 03 February 2019

Subscriber: Mount Sinai School of Medicine; date: 03 February 2019

You might also like