You are on page 1of 6

Measuring Musical Consonance and Dissonance

Michael C. LoPresto

Citation: The Physics Teacher 53, 225 (2015); doi: 10.1119/1.4914564


View online: https://doi.org/10.1119/1.4914564
View Table of Contents: https://aapt.scitation.org/toc/pte/53/4
Published by the American Association of Physics Teachers

ARTICLES YOU MAY BE INTERESTED IN

Teaching Physics with Music


The Physics Teacher 53, 415 (2015); https://doi.org/10.1119/1.4931010

The physics of musical scales: Theory and experiment


American Journal of Physics 83, 835 (2015); https://doi.org/10.1119/1.4926956

Tonal Consonance and Critical Bandwidth


The Journal of the Acoustical Society of America 38, 548 (1965); https://doi.org/10.1121/1.1909741

Clash of Harmonics in Stravinsky’s The Rite of Spring


The Physics Teacher 56, 218 (2018); https://doi.org/10.1119/1.5028234

A Bit of Quantum Mechanics


The Physics Teacher 53, 230 (2015); https://doi.org/10.1119/1.4914565

Chladni Patterns on Drumheads: A “Physics of Music” Experiment


The Physics Teacher 49, 24 (2011); https://doi.org/10.1119/1.3527750
Measuring Musical Consonance and
Dissonance
Michael C. LoPresto, Henry Ford College, Dearborn, MI

M
ost combinations of musical tones are perceived Table II. Frequency ratios and intervals of a chromatic scale.
as either consonant, “pleasing” to the human ear, Musical Musical interval Just frequency
or dissonant, which is “not pleasing.” Despite note ratio = N/M
being largely subjective in nature, sensations of consonance C Unison 1/1
and dissonance can be quantified and then compared to the
C# Minor second 16/15
judgments of human subjects. The following is a description
of several simple studies that can be carried out in the class- D Major second 9/8
room of a physics of music or science of sound course. D# Minor third 6/5
E Major third 5/4
Background F Fourth 4/3
A complex musical tone is a combination of a fundamental
Augmented fourth/ 45/32
frequency and higher harmonics of frequencies that are inte- F#
Diminished fifth
ger multiples of the fundamental. A simple musical tone, like
G Fifth 3/2
the sound of a vibrating tuning fork, consists of only a funda-
mental, generating a simple waveform that will be perceived G# Minor sixth 8/5
by the ear as just that, very “simple” or “pure.” The presence A Major sixth 5/3
of the higher harmonics creates a more complex waveform A# Minor seventh 16/9
causing the ear drum to vibrate simultaneously at multiple B Major seventh 15/8
frequencies, creating a sensation of a tone with more musical
C' Octave 2/1
sound quality or, in musical terms, timbre.
Two musical tones being sounded simultaneously is known The chromatic scale, shown in Table II, fills in the gaps. Every
as a dyad. The ratio of the fundamental frequencies of the note is an interval of a half step above the previous one. The
tones determines at which step or degree in a musical scale chromatic scale2 also contains minor intervals that have a
they will occur and thus the musical interval between them. slightly smaller frequency ratio with the bottom note than the
major intervals of the same name. A diminished interval is a
Table I. Frequency ratios of a musical major scale using the musical
intervals of just intonation. half step less than a perfect interval, an augmented interval
is a half step more. Since it is three full musical steps or tones
Musical C D E F G A B C'
above the bottom note of a scale, the interval of an augmented
note
fourth, or diminished fifth, is usually referred to by musicians
Frequency 1/1 9/8 5/4 4/3 3/2 5/3 15/8 2/1
as a tritone.
ratio
When two complex tones are sounded simultaneously and
Interval Unison Major Major Perfect Perfect Major Major Per-
the frequencies of adjacent harmonics are within less than a
name sec- third fourth fifth sixth sev- fect
ond enth octave
ratio of 6/5, an interval of a minor third, as seen in Table II, a
“roughness” is perceived that is caused by beats between the
two frequencies. This is the cause of the sensation of musical
Table I shows the frequency ratios of the musical intervals of dissonance, an unpleasantness perceived by the human ear.
just intonation.1 These are the ratios that the frequencies of The opposite sensation, musical consonance, is generally con-
the tuning forks in an ordinary laboratory set are above that sidered to be the absence of dissonance.3
of the C-256 Hz fork. Each interval is named after the order in Dissonance between two pure musical tones (again, tones
which it appears in the scale. All the intervals in just intona- consisting only of a fundamental frequency) occurs when
tion are called major intervals except the fourth, the fifth, and the frequencies are within a critical band in which the ear
eighth (the octave), which are referred to as perfect intervals. has difficulty distinguishing between them. Through much
A major scale, shown in Table I, is composed of all major of the audible range, the critical bandwidth is about 1/3 of a
and perfect intervals and has eight steps. The interval that musical octave or four semitones, a frequency ratio of about
each note makes with the one preceding it in the scale is not 1.26, a musical interval of about a major third. Maximum dis-
the same. The intervals between the third and fourth and sev- sonance occurs at about one-fourth of the critical bandwidth.
enth and eighth steps are a musical half step, or semitone, the If the frequencies are a major third or farther apart, then little
smallest interval used in Western music. The ratios between or no dissonance is perceived. If frequencies are less than a
all the others are a full musical step or tone, two half steps. just noticeable difference (JND) apart, the beats are washed

DOI: 10.1119/1.4914564 The Physics Teacher ◆ Vol. 53, April 2015 225
out and the tones fuse together into a single pitch, or a unison,
from which no dissonance is heard.4-6 Class dissonance ratings
25
Beats are readily demonstrated with two tuning forks of N = 18

Class dissonance rating


N = 17
the same frequency mounted on resonance boxes and sound- 20

ed with one having a rubber band or small clamp on one of 15


the prongs to slightly lower its frequency. A demonstration
of this type (or one using electronic frequency generators) 10

should accompany any discussion of consonance and disso- 5


nance.
0
Figure 1 shows the first six harmonics of two complex mu- unison m2 M2 m3 M3 P4 tritone P5 m6 M6 m7 M7 octave
sical tones of an interval of a perfect fifth, N/M = 3/2, a minor Musical interval
sixth, N/M = 8/5, and a major second, N/M= 9/8. For exam-
ple, since the frequency ratio of an interval of a perfect fifth is Fig. 2. Dissonance ratings (normalized) of each interval as rated
by two different sections of a science of sound and light class.
N/M=3/2, the harmonics of the lower pitch are represented as
2, 4, 6… , and the harmonics of the higher pitch as 3, 6, 9… .
As mentioned above, dissonances (marked in red in Fig. 1) is the dissonance or dissonance rating of each interval rela-
will be perceived when the frequency ratio between adjacent tive to the others. Figure 2 shows data taken in this manner
harmonics is less than 6/5. The major second has the most ad- in two different sections of a science of sound and light class.
jacent harmonics with frequency ratios within this range and The data are normalized for comparison by dividing the dis-
is therefore the most dissonant of the intervals shown, while sonance rating for each interval by the total of the dissonance
the perfect fifth can be seen to be the least dissonant. ratings of all the intervals in the entire octave.
Note in Fig. 1 that the larger the numbers in the frequency Data taken in a classroom can be used in several ways.
ratio, the more dissonances there are, while there are less in a One is to simply compare the class dissonance rating of each
tone with smaller numbers in its ratio. This holds true in gen- interval to see if student perceptions of the intervals agree
eral; musical intervals with smaller whole numbers in their with which intervals are generally considered members of
frequency ratios are perceived as the more consonant, while the categories perfect consonances, imperfect consonances, or
those with ratios involving larger numbers are perceived as dissonances. As mentioned above, intervals with small num-
more dissonant. ber frequency ratios are considered consonances and those
with larger or nonconsecutive numbers in their ratios are the
Taking data and making simple compari- imperfect consonances, while those with the largest numbers
sons in their ratios are considered dissonances.7 Table III shows
Data with human subjects can be taken by playing in turn the rated dissonance of the intervals by the two class sections
each of the 12 musical intervals in the chromatic scale in a compared to these classifications.
random order on a piano or portable electronic keyboard Another useful comparison is to rate the intervals accord-
while asking the subjects simply to indicate whether or not ing to their order of merit. This is simply a list of the intervals
they “liked” what they heard by checking “yes” or “no” on a ranked in order of decreasing consonance or increasing
form provided for them. The total number of “no” responses dissonance based on the combinations of ratings of smooth-
ness, purity, blending, and fusion. Intervals with less beating
between harmonics are considered smooth. Purity is rated by
the number of frequencies present that are harmonics of the
Table III. Dissonance ratings of musical intervals from class
sections of N=18 and N= 17 students compared to how they are
generally perceived.
Musical Frequency Class Class
interval ratio dissonance dissonance
rating rating
(N = 18) (N = 17)
Perfect Unison, 1/1, 2/1, 0,1,0,1 0, 3, 0, 1
consonances octave, P5, 3/2, 4/3
P4
Imperfect M3, M6, 5/4, 5/3, 0,1,4,5 1, 0, 6, 8
Fig. 1. The first six harmonics of two complex musical tones of an consonances m3, m6 6/5, 8/5
interval of a perfect fifth, N/M = 3/2, a minor sixth, N/M = 8/5, and a
Dissonances M2, M7, 9/8, 15/8, 13, 16, 16, 13, 10, 17,
major second, N/M =9/8. Adjacent harmonics within a frequency
m2, m7, 16/15, 7, 14 8, 12
ratio of less than 6/5, those that will cause dissonance to be per-
ceived, are marked in red. tritone 16/9, 45/32

226 The Physics Teacher ◆ Vol. 53, April 2015


Table IV. Class dissonance ratings compared to the order of merit.

Interval octave P5 M6 M3 P4 m6 m3 tritone m7 M2 M7 m2

Order of merit 1 2 3 4 5 6 7 8 9 10 11 12
Class data 1 0 1 0 1 5 4 14 7 13 16 16
(N = 18)
Class data 3 0 0 1 1 8 6 12 8 13 10 17
(N = 17)
Table V. Just frequency ratios and their decimal equivalents compared to the frequency ratios of the equal tempered
scale. The highlighted ideal ratios have different equal tempered values than in just intonation.

Musical note Musical interval Just ideal ratio n Equal tempered Equal tempered
decimal equivalent ideal ratio (if
different than just)
C Unison 1/1=1.000 0 2n/12 = 1.000
C# Minor second 16/15<1.067 1 1.059 18/17<1.059
D Major second 9/8=1.125 2 1.122
D# Minor third 6/5=1.200 3 1.189
E Major third 5/4=1.250 4 1.260
F Fourth 4/3<1.333 5 1.335
F# Tritone 45/32<1.406 6 1.414
1.414 17/12<1.417
G Fifth 3/2=1.500 7 1.498
G# Minor sixth 8/5=1.600 8 1.587
A Major sixth 5/3<1.667 9 1.682
A# Minor seventh 16/9<1.778 10 1.782
B Major seventh 15/8=1.875 11 1.888
1.888 17/9 <1.889
C' Octave 2/1=2.000 12 2.000

fundamentals of both tones in the interval and blending is a tempered scale is the intervals between all adjacent notes in
more subjective sense of whether the notes seem to “belong” the chromatic scale being precisely equal. This is achieved
together. Fusion is the sense of whether or not the two tones by the frequency ratio of each successive note in the scale, n,
combine to sound as one.8,9 Table IV shows the dissonance having a frequency that is a factor of 2n/12 above the bottom
ratings from the class sections compared to the order of (n = 0) frequency, where n = 1, 2, 3… to 12, as seen in Table
merit. V. 14
Most of the ideal ratios for the just intervals are approxi-
A dissonance metric mately the same as those for the equal tempered intervals, but
A more quantitative comparison can be done using a dis- there are in fact a few (those highlighted in Table V) equal
sonance metric. One possible dissonance metric is the sum10 tempered ideal ratios that are different. Recalling that the
of the numbers in a musical interval’s ideal ratio.11 The ideal ideal ratio for an interval is the fraction that has the closest
ratio for an interval is the fraction N/M that has the closest decimal equivalent to the frequency ratio of the interval, it is
decimal equivalent to the frequency ratio of the interval. necessary to have the correct ideal ratios to calculate accurate
The ideal ratios for the just intervals are the frequency ratios dissonance metrics for comparison with data taken with hu-
shown in Tables II and III. The sum dissonance metric for an man subjects.
interval is N +M. As seen in Table V, there are three intervals in the equal
The physical basis for the sum dissonance metric is that tempered scale that have different ideal ratios than their
intervals with small whole number ratios, those having a just counterparts. They are the following: the minor second
smaller sum, will have more coinciding upper harmonics and 18/17<1.059 is a better match to the equal tempered semi-
therefore less dissonance.12, 13 tone (2)1/12<1.059 than the just 16/15<1.067; the tritone
Calculating the sum dissonance metric for the just inter- 17/12<1.417 matches the equal tempered (2)6/12 = (2)1/2 <
vals is simple, but when data are taken with human subjects 1.414 better than the just 45/32<1.406; and the major sev-
using a piano or keyboard, the musical intervals heard by the enth 17/9<1.889 is just slightly closer to the equal tempered
subjects are not those of just intonation, but rather those of (2)11/12 <1.887 than the just 15/8=1.875.
the equal tempered scale, the intonation that is currently used The reason for using the equal tempered scale in Western
in Western music. As the name suggests, the basis of the equal music is so musical pieces can be played in more than one

The Physics Teacher ◆ Vol. 53, April 2015 227


musical key. Playing in one musical key means using the
notes of a musical scale starting on a specific note. Class rating and sum metric
Changing keys means changing the notes used to those in a 25
N = 18
scale starting on a different note. This can be problematic in 20
sum metric

Dissonance rating
just intonation because not all the half steps or semitones be- 15
tween each step of the scale are the same. For instance, as can
10
be seen in Table II, the fourth note in the scale is a frequency
5
ratio of (6/5)/(9/8)=(6/5)*(8/9)=48/45=16/15<1.067
above the third note, while the fifth note is a ratio of 0
unison m2 M2 m3 M3 P4 tritone P5 m6 M6 m7 M7 octave
(5/4)/(6/5)=(5/4)*(5/6)= 25/24<1.042 above the fourth note.
Musical interval
In equal tempered tuning, each note in a scale is exactly a fre-
quency ratio of 21/12 <1.059 above the previous one, keeping Fig. 3. Class dissonance ratings compared to the sum dissonance
all half steps the same. metric.
Close examination of Table V shows that the decimal N=17 class rating vs sum metric
equivalents of the just frequency ratios for the perfect conso- 25
y = 1.1405x - 1.0805
nances (the octave, perfect fourth, and perfect fifth) and the R2 = 0.79373

Class dissonance rating


20
imperfect consonances (the major and minor third and the
major and minor sixth, as defined in Table III) are very close 15

to the equal tempered ratios. This means that the frequency 10


ratios for all these intervals, even in equal tempered tuning,
are still very close to being ratios between small whole num- 5

bers and will therefore still be perceived as consonant.


0
Figure 3 is a normalized plot of the class dissonance rat- 0 2 4 6 8 10 12 14 16 18 20
Sum metric
ings (N=17) compared to the sum dissonance metric calcu-
lated with the equal tempered ideal ratios. Fig. 4. Normalized plot of class dissonance rating vs the sum dis-
sonance metric.
Data taken with human subjects can also be compared to
Dissonance rating and sum metric
a dissonance metric with a normalized plot of the dissonance 20
ratings versus a dissonance metric. Figure 4 shows such a plot 18
N = 100
sum metric
for the N=17 class dissonance ratings compared to the sum 16
dissonance metric. 14
Clearly N=17 is a very small data set for drawing con-
Dissonance rating

12

clusions. This can be impressed upon students by showing 10

results from larger data sets. Figures 5 and 6 show data taken 8

for N=100 subjects in several general education science sec- 6

tions in previous semesters as advertisements for the sound 4

and light course. It should be noted that a number of students 2


0
from these classes went on to take the sound and light class, unison m2 M2 m3 M3 P4 tritone P5 m6 M6 m7 M7 octave

suggesting that the advertising was successful. Musical interval


Fig. 5. Dissonance ratings of N=100 human subjects compared
Musically and non-musically trained to the sum dissonance metric.
subjects N = 100 dissonance rating vs sum metric
N=49 human subjects with musical training were sur- 20
veyed as part of an advertisement for the sound and light 18

course during a concert choir rehearsal. This provided the 16


y = 0.9654x + 0.2664
Dissonance rating

14
opportunity to compare the dissonance ratings by these sub- 12
R2 = 0.80689

jects to those of the N =100 subjects without musical training 10


who were surveyed in the general education science classes. 8
The results, shown in Fig. 7, indicate that musically 6

trained subjects disliked the minor seventh and the major 4


2
seventh, two intervals generally considered as dissonant 0
(see Table III), more than the not musically trained subjects. 0 2 4 6 8 10 12 14 16 18 20
Those not musically trained disliked the major and minor Sum metric

sixth, considered imperfect consonances (again see Table III), Fig. 6. Normalized plot of dissonance ratings of N = 100 human
more than the musically trained subjects. subjects vs the sum dissonance metric.

228 The Physics Teacher ◆ Vol. 53, April 2015


Acknowledgments
Musical and Non-musical
25 I thank all HFC students who participated in data collection
Music and HFC Director of Choirs, G. Kevin Dewey, for allowing
20 Non-music data to be taken during concert choir rehearsal time and for
Dissonance rating

informal discussions about the subject of this paper. I also


15
thank the anonymous TPT referees for the time and effort
10 put into several reviews that provided many helpful and
detailed suggestions on how to improve this manuscript.
5

0
References
unison m2 M2 m3 M3 P4 tritone P5 m6 M6 m7 M7 octave 1. L. Gunther, The Physics of Music and Color (Springer, New
Musical interval York, 2012), pp. 363–365.
Fig. 7. Comparison of dissonance ratings of the intervals as 2. Ref. 1, pp. 365-366.
judged by human subjects with and without musical training. 3. H. Helmholtz, On the Sensations of Tone (Dover, New York,
1954), p. 194.
Possible reasons for these differences in perception are 4. R. Plomp and J. M. Levelt “Tonal consonance and critical band-
width,” J. Acoust. Soc. Am. 38, 548–560 (1965).
that the musically trained subjects simply hear the different
5. G. Loy, Musimathics–The Mathematical Foundations of Music,
intervals more keenly because of their training or that, in
Vol. 1 (MIT Press, Cambridge, MA, 2006), pp. 178–186.
some cases, they recognize the intervals and know how they
6. D. J. Benson, Music: A Mathematical Offering (University Press,
are “supposed” to respond. This is a reason that some suggest Cambridge, MA, 2007), pp. 150–153.
they should not be used as subjects in studies of this type.15 7. J. G. Roederer, The Physics and Psychophysics of Music-An In-
It is also possible that the musically trained subjects are more troduction, 4th ed. (Springer, New York, 2008), p. 171.
familiar with the use of the imperfect consonances in music 8. H. F. Olsen, Music, Physics and Engineering, 2nd ed. (Dover,
and thus find them less dissonant than the non-musically New York, 1967), p. 260.
trained subjects. These results and possible explanations were 9. C. E. Seashore, Psychology of Music (Dover, New York, 1967),
also shared and discussed with students in the sound and pp. 131–133.
light class. 10. Ref. 5, pp. 56–60.
11. J. Pierce, “Consonance and Scales,” in Music, Cognition and
Conclusion Computerized Sound: An Introduction to Psychoacoustics, ed-
All of the exercises described can be comfortably run ited by P. R. Cook (MIT Press, Cambridge, MA, 2006), p. 175.
during a normal laboratory period. Data can be collected in 12. Ref. 7.
class in the manner described above, then by show of hands 13. Ref. 5, pp. 57–58.
the number of students rating each interval as dissonant can 14. Ref. 1, pp. 367–370.
be determined. The students can then use the class data to 15. W. A. Sethares, Tuning, Timbre, Spectrum, Scale, 2nd ed.
(Springer, London, 2005), p. 92.
determine how well their dissonance ratings agree with the
classifications in Table III and with the order of merit in Table
Michael C. LoPresto is in his 25th year of teaching physics and
IV. While students are doing this, the instructor can normal- astronomy at HFC; he has been teaching Sound & Light in Fine Arts
ize the data to have it ready for students to make plots similar for the last 10 years and is a regular contributor to TPT.
to Figs. 3 and 4 to compare the class dissonance ratings to the Henry Ford College, Dearborn, MI 48128; lopresto@hfcc.edu
sum dissonance metric. If time is running short, the instruc-
tor can do the plot for the entire class to examine together.
Since one class section is usually a very small data sample,
results could vary from those presented here. However, it has
turned out that in most sections the sum metric has usually
matched the class data well. The small classroom samples
are the reason for showing the results of larger data sets. This
can be informative because, as seen in Figs. 5 and 6, the data
show a higher correlation with the sum dissonance metric,
verifying the in-class results. The use of the larger data sets to
compare the perceptions of subjects with and without musi-
cal training and discussing possible reasons for the differ-
ences is also worthwhile, especially for the numbers of music
students generally present in a science of sound or physics of
music class.

The Physics Teacher ◆ Vol. 53, April 2015 229

You might also like