Harmonic Visualizations of Tonal Music

Craig Stuart Sapp

Center for Computer Assisted Research in the Humanities
Center for Computer Research in Music and Acoustics
Stanford University
email: craig@[Link]

Abstract KS algorithm analyses

BWV notated key prelude fugue
Multi-timescale visualization techniques for displaying the 846 1. C major C major C major
output from key-finding algorithms are presented in this pa- 847 2. C minor C minor C minor
per. The horizontal axis of the key graphs represents time in 848 3. C] major C] major G] major
the score, while the vertical axis represents the duration of 849 4. C] minor C] minor C] minor
an analysis window used to select music for the key-finding 850 5. D major D major D major
algorithm. Each analysis window result is shaded according 851 6. D minor D minor D minor
to the output key’s tonic pitch. The resulting diagrams can be 852 7. E[ major E[ major E[ major
used to compare differences between key-finding algorithms 853 8. E[/D] min. E[ minor D] minor
at different time scales and to view the harmonic structure 854 9. E major E major E major
and relationships between key regions in a musical composi- 855 10. E minor E minor E minor
tion. 856 11. F major D minor F major
857 12. F minor F minor F minor

1 Motivation Table 1: KS algorithm results when applied to entire WTC

book 1 compositions compared to actual tonic keys of the
A tonal composition is usually described as being in a par- music. Identification errors are underlined.
ticular key, such as Brahms’ Symphony No. 4 in E minor, or
Beethoven’s Piano Sonata in F minor, Op. 57; however, rarely
does a piece of music maintain a single key center through- gorithm based on probe-tone ratings generated from experi-
out its entirety. The key of a piece typically starts and ends mental results (Krumhansl 1990). The KS algorithm is im-
the piece, but other key centers are used somewhere in the plemented in the key program contained in the Humdrum
middle of a piece to give form to the music. A simple stereo- Toolkit for musical analysis (Huron). Applying the KS algo-
typical tonal piece might start in the tonic key, modulate to rithm to an entire piece, table 1 lists the analyzed keys for
the dominant, and then return to the tonic key by the end of the first half of the first book of J.S. Bach’s Well-Tempered
the piece. Clavier using the key program. The Well-Tempered Clavier
Currently available key-finding algorithms are not very is an excellent source of test material for testing key-finding
sensitive to identifying modulations, and if given a selection algorithms, because each set of prelude and fugue in the col-
of music which contains several modulations, the algorithms lection are in a different key, starting in C major and then pro-
can only identify what the most likely key is for the entire gressing chromatically through all 24 major and minor keys.
selection. If there are two key areas in a selection of mu- Table 1 points out two common errors generated by key-
sic, then the algorithm hopefully assigns the stronger key the finding algorithms in general. The first error is in fugue no. 3
best score while the other key area hopefully is assigned the where the KS algorithm identifies the dominant rather than
second best score. A key identification error may occur if the correct tonic key of C] major. The second error occurs
an entire piece of music is presented to a key-finding algo- in the eleventh prelude where the relative minor is identified
rithm, since the secondary key areas may overpower the start- rather than the key of F major.
ing/ending key or bias the primary key towards a closely re- These two errors are primarily due to more than one key-
lated key, such as the dominant or relative minor key areas. area being present in the analyzed music, causing slight off-
One of the best algorithms for determining the key in a sets in the tonic key weightings such that a closely related key
region of music is the Krumhansl-Schmuckler key-finding al- becomes more likely for the algorithm than the actual key.
A fifth-relation error can occur if the secondary key areas are 2 Diagram Types
predominantly all above or below the tonic key in the circle
of fifths. The second error is a modality error. The correct 2.1 Key-To-Color Mappings
key signature was identified, but the distribution of notes in
the music was such that the tonic was incorrectly identified. To display data from key-finding algorithms in a compact
Of course, the key identification errors in the two com- visual manner, each key is mapped to a different color. The
positions from the Well-Tempered Clavier could be fixed by principle key-to-color mapping being used is shown in ta-
only applying the algorithm to the first and last parts of the ble 2. The colors of the rainbow are mapped onto the circle
music, since these sections are more likely to contain the tonic of fifths collapsed to the seven diatonic pitches. For example,
key. However, side-stepping the issue in this manner creates the key C is assigned the color green. Ascending the circle
other problems: (1) How much of the beginning and ending of fifths yields G (blue), D (indigo blue), A(purple), E (red).
of the piece should be examined? and (2) What if the compo- Going the opposite direction in the circle of fifths takes you
through the rainbow colors in the opposite direction: F (yel-
low), B[ (orange), E[ (red).
sition starts in one key and ends in another? The true problem
to solve is how to identify correctly regions of stable key cen-
ters and regions of modulation in a piece of music. diatonic diatonic
If a more detailed view of a pieces’s key structure is de- pitch class color pitch class color
sired beyond the “key” of the piece, then a moment-by-moment E red G blue
view of the key centers in the piece will give a much more de- B orange D indigo
tailed description of the piece. Krumhansl proposes applying F yellow A violet
a sliding analysis window to the notes in a piece to generate C green
localized key measurements. This gives a good overview of
the key relationships in the music, but results can be sensitive root (R, G, B) root (R, G, B)
to the analysis window size. A problem with this sliding win- C[ ( 36, 255, 0) G[ ( 54, 200, 218)
dow technique is that the global importance of a local key is C ( 0, 255, 0) G ( 63, 191, 255)
not apparent in the local context. A music theorists would as- C] ( 9, 246, 36) G] ( 63, 177, 255)
sign a sequence of key centers to a piece of music using both D[ ( 63, 109, 255) A[ (118, 41, 255)
the global and local characteristics of the music. If a region D ( 63, 95, 255) A (127, 31, 255)
is difficult to assign a key label, then music outside the local D] ( 73, 86, 255) A] (145, 27, 219)
region will be considered. It is very difficult to have a com- E[ (237, 4, 36) B[ (255, 109, 0)
puter generate a reasonably accurate sequence of musically E (255, 0, 0) B (255, 127, 0)
relevant key centers. E] (255, 18, 0) B] (255, 145, 0)
Key visualization techniques described in the following F[ (255, 237, 0) F] (218, 255, 0)
section avoid the problem of choosing a fixed analysis win- F (255, 255, 0)
dow duration by instead using all possible analysis window Table 2: Sample RGB color mappings for key tonics.
durations. The interpretation of the local key problem is par-
tially solved with these display methods because the height
to which a key region survives in a diagram demonstrates the This color mapping is designed to be intuitively easy to
relative strength of that key region. Strong modulations are navigate, since closely related keys map into closely related
represented by large vertical structures, while tonicizations colors. The only drawback of this mapping is that there are
are represented by smaller vertical structures. only 7 colors while there are 12 (or more) pitch classes—
It is also possible to view the behavior between various notice that E and E[ have the same red color. Since enhar-
key-finding algorithms at different analysis window sizes and monic keys are located far apart in key space, this mapping
to see how they interact around regions of modulation. Key- is unambiguous for most tonal music and should only cause
algorithms can then be compared using the diagram methods confusion in rare cases. A monochromatic color mapping has
below to see how the different algorithms handle the same been used in this paper for printing in black-and-white, but
music at various time scales. Numerous computational algo- it is possible to view example figures from this paper on the
rithms for identifying keys using computers have been pro- World Wide Web (Sapp).
posed since the 1960’s. See recent work and surveys on key- Of course, the choice of a color mapping from key to color
finding algorithms by Temperley (1997), Chew (2000) and is arbitrary, and different color mappings may be suitable for
Shmulevich and Yli-Harja (2000). Sleator and Temperley different types of music. For tonal music, the relationship
make the source code for their harmonic analysis programs between keys is predominantly based on the circle of fifths;
available on the web (Sleator and Temperley). therefore, arranging the circle of fifths onto the rainbow is ad-
vantageous. Many other types of color mappings are possible,
including:
  indicating major/minor modalities with brightness. For If the N levels of the vertical axis are displayed with
example C major could be light green and C minor equal heights, the lower levels overpower the higher levels
could be dark green. as demonstrated on the left side of figure 2. For example, the
ratio between the first and second levels at the top of the dia-
 indicating sharp keys brighter, and flat keys darker to gram is a ratio of 1/2 or a 50% decrease in window duration.
unambiguous enharmonic keys, for example C-flat could The ratio between the 100th and 101st levels is 100/101 or a
be dark green and C-sharp could be light green. 1% decrease in duration. The higher the number of divisions,
 using hue instead of the rainbow to maximize distance the less change there is between analysis window sizes, and
between keys in color space. therefore less importance should be given to the visual impact
between levels.
 using a monochromatic mapping for optimal display in
black and white (as well as for color-blind people).
 displaying music in the key colors of its synesthetic
composer.
It is also possible to label the tonic of a composition in a cer-
tain color, and then label all other key areas in relation to the
tonic. But for this system to work, you would need to guar-
antee that the correct key has been chosen as the tonic key. Figure 2: Type 1 plot analysis window layout. Left: linear
vertical scale; Right: logarithmic vertical scale.
2.2 Type 1: discrete time/discrete roots
To correct for the emphasis on larger division levels, the
The first type of diagram for key display divides a piece vertical axis can be adjusted with a logarithmic scaling. A
of music into successively smaller analysis window units as strong logarithmic scaling weights the upper levels of the dia-
outlined in figure 1. The top level of the diagram analyzes all grams too much, so good scalings which equally emphasize
of the notes in the selection to generate a key identification. the top level and bottom levels of the diagrams is generated
The second level of the diagram then splits the music into two by this example scaling:
equal parts and generates a key for the music in each half, and  n + 1) 
so on. Each successively lower level of the diagram divides m=M 1
log(
log(N + 1)
the music into a greater number of equal-sized analysis win-
Where N is the total number of analysis window levels, n is
dows which get continually smaller in duration.
the current level indexed from 0, M is the number of vertical
pixels in the picture, and m is y-axis maximum pixel location
of level n with respect to the bottom of the picture. The right
side of figure 2 displays this type of mapping where the areas
of the window blocks are closer in proportion to the change
in window duration between levels.
Figure 3 displays a type 1 key diagram for the first move-
ment of Mozart’s Viennese Sonatina No. 1 in C major. The
horizontal axis displays the music from the start (left) to the
ending (right) of the piece. The vertical axis displays the anal-
ysis results for small window durations at the bottom of the
Figure 1: Type 1 analysis window configuration. picture and larger analysis window durations at the top of the
picture. Pitch labels identify important key/chord regions in
The lowest level will then divide the musical selection the piece. Notice that the key of C forms the dominant shaded
into N parts. The duration of the analysis windows in this area of the picture which coincides with the identification of
level are total-time/N time units wide. Typically N is set to this piece being in the key of C major.
the number of beats in the music, but any smaller or larger The bottom portion of fig. 3 represents individual chords.
value of N can be used as well. When using a value of N As the analysis windows get larger towards the top of the pic-
equal to the value of beats in the music, the bottom level of ture, the chords merge with adjacent chords. Some chords get
the diagram displays chord roots, and the top levels represent absorbed quickly into adjacent chords, particularly if they are
the strong key areas present in the piece. remotely related to the surrounding chords. By the middle of
the vertical axis, the strongest chords start to represent key
 analysis window size
Figure 3: Mozart Viennese Sonatina No.1 in C Major,
start (of piece) end (of piece)
Mvmt.1: type 1 plot with logarithmic vertical scale. time

Figure 4: Type 2 analysis window arrangement.

areas. Some key areas have a short duration and are absorbed
into stronger keys areas higher up in the diagram. The top
portion of the figure shows the major key regions of the piece dow duration, the analysis window is slid continuously over
which are primarily the keys of C major and G major. the entire piece of music, placing the result of the key-finding
The sonatina movement is clearly in C major, since the algorithm at the center point of the analysis window. Typi-
upper left and right portions of the diagram are in C. The fig- cally, the analysis windows of the lowest level in the diagram
ure also shows a strong region of G major in the 2nd quarter contains one beat of the music. The placement of the analysis
of the piece. There is a distinction between the F modulation output at the center of the window generates a characteristic
near the middle of the piece and the temporary modulation triangular shape which makes the diagram easy to distinguish
(tonicization) of F closer to the end of the piece. Toniciza- from type 1 plots.
tions do not extend towards the top of the piece as high as Increased resolution is a big advantage of type 2 plots over
true modulations into a key area. type 1 plots. Figure 5 now shows the Mozart sonatina move-
The striped bands for the key of A in the middle of the ment with a very nicely formed region in the key of G. Also,
movement demonstrate weakness in the particular key-finding the development regions near the center of the movement are
algorithm being used. The A key region is probably a com- easier to distinguish since they are less fragmented.
promise in the algorithm between an overlapping region of
D minor and F major. Another problem is that the F key area
contains a region of C inside. This C region is due to domi-
nant seventh chords in the key of F, so the C regions should
be expected to be smaller and incorporate into F faster than
the plot shows.

2.3 Type 2: continuous time/discrete roots

The primary drawback of type 1 key diagrams is that quan-
tization errors increase in the higher analysis window levels
where there are fewer divisions to represent the entire piece.
Therefore, a second type of diagram is presented in this sec-
tion which gives equal resolution at all time scales. Instead
of coloring the entire analysis window duration with the key Figure 5: Mozart sonatina: type 2 plot with linear vertical
color, a single pixel centered in the middle of the analysis scale (compare to fig. 3).
window is drawn. Note that computation time for type 2 plots
is about 30 to 50 times greater that for type 1 plots. Just as there was a problem with the vertical scale in type 1
Figure 4 gives an overview of the windowing method used plots which show too much of the lower time-scales, figure 5
to create type 2 key diagrams. The top level of the diagram displays too much of the higher time-scales. Note that the
contains an analysis window that can hold the entire piece of second level of a type 1 plot is now found midway on the ver-
music. For that window, a single pixel is displayed at the top tical axis in a type 2 plot. The unaltered form of the type 2
of the diagram. To generate the lower parts of the diagram, plots is equivalent to a strong logarithmic scaling of the type 1
the analysis windows continually get smaller. For each win-
Figure 6: Mozart sonatina: type 2 plot with logarithmic ver-
tical scale (compare with fig. 5.) The vertical scale is approx-
imately the same as in fig. 3.

Figure 7: Continuous nature of some key-finding algorithms’

plots. Therefore, a more balanced view of the top and bottom
key scores. Bottom part of figure is a midrange zoom-in on
levels can be achieved by scaling the vertical axes logarithmi-
fig. 6. Top part of figure shows the continuous key-finding
cally, as shown in figure 6.
scores for important key regions highlighted in the lower fig-
ure.
2.4 Type 3: continuous time and roots
Most key-finding algorithms calculate the likelihood of given moment in the music is plotted. Type 1 and 2 key dia-
every possible key and then assign the key with the high- grams will only display the most likely key at any given mo-
est score as the best key to fit the music. For example, Ta- ment, which in this case is the sequence: C, G, A, F, and then
ble 3lists the KS algorithm scores of each possible key for a return to C. What is lacking in these visualizations is an
the entire Mozart sonatina movement. The third category of indication of how certain the algorithm is in its choice of key.
key diagram described in this paper takes these secondary key Two measurements in specific can be extracted from in-
probabilities into account to generate color-interpolated pic- formation shown in figure 7 to give a more continuous key
tures of key based on the general form of type 2 plots. diagram. (1) Clarity: The higher the score, the better the hy-
pothesis key fits the music. Regions of musical stability are
r-value key r-value key likely places to find a clear key center. For example, the reca-
0.945 C major -0.068 E[ major pitulation into C major in the first movement of the sonatina
0.770 G major -0.158 E major coincides with the highest score for C major. (2) Ambiguity:
0.665 E minor -0.160 A major regions of development such as in the middle of the sonatina
0.481 C minor -0.260 G] major consist of closely scored keys. The algorithm has difficulty in
0.479 A minor -0.321 C] minor choosing the best key in this case, because modulations hap-
0.417 F major -0.415 B major pen so quickly compared to the more stable exposition and
0.403 G minor -0.420 G] minor recapitulation. Key identification errors are more likely in re-
0.163 D minor -0.436 F] minor gions of high ambiguity.
0.110 D major -0.450 B[ minor Experiments in displaying just the clarity or ambiguity
0.063 F minor -0.557 C] major yield promising diagrams which might prove useful in mu-
0.027 B[ major -0.646 D] minor sic analysis; however, key scores are difficult to interpret and
0.019 B minor -0.650 F] major relate to each other over different time scales. Figure 8 dis-
plays a plot of clarity between the best and second best keys.
Table 3: KS algorithm r-value scores of each possible key for Notice that the short development region in the middle of the
the Mozart sonatina, sorted from most likely (highest r-value) movement is clearly indicated by the dark band rising verti-
to least likely key. cally in the center of the plot. Plots of ambiguity look similar
to the regular best-key plots because they usually outline the
Figure 7 plots algorithmic score weightings of the four borders between key areas.
primary key centers taken from the midsection of figure 6. Interpolating the colors of the best and second best keys
In this figure the relative strengths of each key score at any
 into the primary key of C major.
The interpolation for figure 10 was done linearly in the
RGB color space. Interpolation in another color space, such
as HSI (Hue-Saturation-Intensity) may be possible, but pro-
ducing visually pleasant diagrams with this type of interpola-
tion has been difficult to control. An HSI interpolation could
be used to distinguish between modulations by fifths from
other types of more distant key modulations. The more dis-
tant the modulation, the higher would be the saturation of the
colors in the modulation region.

Figure 8: Plot of clarity – light regions indicate higher cer-

tainty of correctness by the key-finding algorithm and darker
regions indicate little difference in the scores between best
and second best keys.

gives a nice continuous diagram. Figure 9 shows a plot of the

second best keys for the sonatina. Notice that in this piece,
the second best key is usually tonic in a dominant section,
and the dominant in a tonic section, although there is a strong
tendency towards the subdominant during the beginning of
the piece and close to the final cadence.

Figure 10: Mozart sonatina: type 3 plot interpolating between

best root (fig. 6) and second best root (fig. 9).

3 Applications
3.1 Comparing key-finding algorithms
The key diagrams described in the previous section were
developed for two purposes: (1) examine the interaction be-
tween key regions in a piece of music, and (2) compare the be-
havior of different key-finding algorithms. Most of the plots
in this paper use an algorithm I am currently developing for
finding the roots of chords. This algorithm matches the output
Figure 9: Mozart sonatina: second-best keys. Compare to
of the KS algorithm for the most part, although the calcula-
fig. 6 which shows the best roots.
tions to derive the analysis are completely different.
Figure 11 shows a type 2 plot of the Mozart sonatina
Figure 10 is an interpolation between the best and second
movement which can be compared directly to figure 6. Good
best keys at each point in figures 6 and 9. For each analysis
features: (1) definition of C major at start of piece is clearer
window size, the maximum clarity between the best and sec-
than in figure 6, (2) dominant key areas in the piece are C and G,
ond best score is used to normalize the interpolation: at the
(3) incorrect A key region in middle of the piece is less pro-
point of maximum score separation, the best key is displayed
nounced, (4) more solid F key region in middle of the piece,
fully in its own color. At points during the piece where the
(5) nicely behaved boundary between G and C key areas.
best and second best keys trade places (at 100% ambiguity),
Not-so-good features: (1) E minor key tendencies near end
the color assigned to the diagram is halfway between the col-
of the movement (modal error from G major), and (2) incor-
ors of the two key centers. Notice that the higher levels of
rect identification of key regions just before F major area.
secondary key regions (G and F) in figure 10 blend smoothly
 strengths and weakness in a particular algorithm are much
easier to detect, since a large number of analyses are needed
to make a single picture. In particular, improvements to the
handling of modulation areas should yield better boundaries
between key-regions.

3.2 Music Analysis

The key diagrams presented in this paper have numerous
potential applications in harmonic analysis. In music theory
training, the visualization maps of the harmony can assist stu-
dents in understanding musical structures which are difficult
to explain. For example, the diagrams could help in under-
standing the difference between a modulation and a toniciza-
tion. Also, the plots can be an objective tool to explain the
Figure 11: Mozart sonatina: using the KS algorithm. (com- conflicting interpretations of music theorists.
pare to fig. 6). Another type of music analysis application may be the
identification of harmonic form, and style. Highly tonal mu-
sic such as the Mozart sonatina, used as an example in this pa-
A slightly stronger emphasis on the dominant key can also per, form clearly defined high-level key structures in the dia-
be seen in figure 11 as compared to figure 6. The large G ma- grams. Baroque and Romantic era music generally contain
jor structure proceeds higher up in the diagram than it does in more elaborate key relations and therefore have a tendency
figure 6 and the fragmented G region around the 2/3 point in to contain more detail in the higher levels of the diagrams as
figure 11 is much less pronounced in figure 6. Thus, it can be compared to Classical music.
inferred that the KS algorithm will be more prone to identi- The diagrams for J.S. Bach’s prelude and fugue in C mi-
fying the dominant key of the correct key than the algorithm nor from WTC 1, displays an interesting contrast. The pre-
used to generate figure 6. lude’s diagram contains smooth key center boundaries at the
Figure 11 can also be used to illustrate the difficulty of higher levels, while the fugue’s diagram contains key centers
extracting a good sequence of modulations in a composition with much more fragmented boundaries. This is to be ex-
as mentioned in the Motivation section. The Mozart sonatina pected since Bach preludes are usually consist of one melody
is a very simple and clear tonal piece. I say that the movement and accompaniment, while fugues are more focused on the
contains a sequence of 5 key regions: (C, G, D, F, C), with the linear aspects of melody rather than on harmony.
following segmentation boundaries:
key area bars length
C major 1 – 15 15 bars
G major 16 – 28 13 bars
D minor 29 – 32 4 bars
F major 33 – 38 6 bars
C major 39 – 63 25 bars
The D minor section is very brief and could be considered
part of the F major region. D minor serves as a transition be-
tween the keys of G and F by switch roles from a dominant
key to a relative minor key. Now notice that this sequence of
keys is not present on any horizontal line in figure 11. There-
fore, no fixed analysis window size in the KS algorithm will
give the correct sequence according to my human-based har- Figure 12: Anton Webern – Op. 27, No. 1. Twelve-tone
monic analysis. Figure 6 fairs a little better with a wide region music (compare to fig. 5 and fig. 13.).
in the middle of the diagram containing the sequence (C, G,
A, F, C), and the algorithm admits that it is having problems As a contrast to tonal music, figures 12 and 13 display
identifying a key in the D minor region (see figure 8). samples of non-tonal music. The Webern piano variation
Further improvements to key-finding algorithms can be shown in figure 12 uses the modern technique found in twelve-
accomplished by using the visualization techniques presented tone music where all twelve pitch classes are used in sequence
in this paper as an evaluation tool. With key diagrams, the
before being repeated. The twelve-tone technique is there-
fore used intentionally to destroy any sense of key through-
out the composition. Notice that the key center boundaries
in figure 12 are fragmented. The diagonal evolution of the
key areas towards the top of the diagram is also unlike the
vertical structures created in the Mozart sonatina. In particu-
lar, the C key region in the middle of the diagram has an in-
verted foundation. In other words, it is difficult to define the
key of C major without playing C major chords. Figure 12
only shows the best keys, when plotting a type 3 diagram of
the twelve-tone piece, the keys areas become even less pro-
nounced than in this type 2 plot.
Figure 13 shows a piece of music from the other side of
music history. This figure displays a type 2 key plot for a Me- Figure 14: Alexander Scriabin – Prelude in E minor, Op.11,
dieval motet from the 13th century—several hundred years No.4.
before the development of functional harmony in the 17th
century. Even though the piece uses a diatonic pitch set, it is
harmonic maps generated from these visualizations of key-
remarkably similar in character to the Webern example. Key
finding algorithms, the interactions and relations between keys
boundaries are fragmented and follow diagonal paths more
become easier to examine.
than vertical paths. Also similar to the Webern example, the
The plots presented in this paper generate fascinating vi-
top level key centers switch between C and D, which are not
sualizations of harmony. Aside from looking nice, they can
closely related keys.
be used to compare the behavior of different key-finding al-
gorithms. Also, they are a good starting point from which to
develop more robust algorithms that can accurately detect key
modulations in a musical sense rather than in a computational
sense.
Many variations on the basic harmonic visualizations pre-
sented in this paper are yet to be explored. For example, all of
the plots presented in this paper use analysis windows which
weight musical pitches in the analysis window equally. Vary-
ing the importance of a pitch inside the analysis windows by
using a triangular or exponential window might yield inter-
esting results.

References
Figure 13: 13th century motet attr. to Petrus de Cruce (com-
pare to fig. 5 and fig. 12. Chew, Elaine (2000). Towards a Mathematical Model of
Tonality. Ph. D. thesis, MIT, Cambridge, Mass.
As a final historical contrast, consider figure 14 which dia- Huron, David. The Humdrum Toolkit: Software for Music
Research. [Link]
grams the key profiles in a prelude written by the late Roman-
[Link]/humdrum.
tic composer Alexander Scriabin. The key boundaries at the
Krumhansl, Carol. (1990). Cognitive Foundations of Musical
higher levels are fragmented, but the primary key is clearly
Pitch. New York: Oxford University Press.
E minor. The top level key centers in this diagram are closely
Sapp, Craig. Tonal Landscape Gallery. [Link]
related (tonic and dominant). The lower level key centers of
A and D are arranged in a vertical manner similar to the key

[Link]/ craig/keyscape.
Shmulevich, Ilya and Olli Yli-Harja (2000). Localized
regions in the Mozart sonatina. key-finding: Algorithms and applications. Music
Perception 17(4), 531–544.
Sleator, Daniel and David Temperley. The Melisma Music
4 Summary Analyzer. [Link]
analysis.
It is more interesting to know the interaction between key
Temperley, David (1997). An algorithm for harmonic analysis.
areas in a piece of music than just knowing the key-label
Music Perception 15(1), 31–68.
for an entire composition or selection of music. With the