COMPUTATIONAL APPROACHES
FOR THE UNDERSTANDING OF MELODY IN CARNATIC MUSIC
Gopala K. Koduri, Marius Miron, Joan Serra` and Xavier Serra
Music Technology Group
Universitat Pompeu Fabra, Barcelona, Spain
gopala.koduri@gmail.com,miron.marius@gmail.com,joan.serraj@upf.edu,xavier.serra@upf.edu
ABSTRACT
The classical music traditions of the Indian subcontinent,
Hindustani and Carnatic, offer an excellent ground on which
to test the limitations of current music information research
approaches. At the same time, studies based on these music
traditions can shed light on how to solve new and complex
music modeling problems. Both traditions have very dis-
tinct characteristics, specially compared with western ones:
they have developed unique instruments, musical forms, per-
formance practices, social uses and context. In this article,
we focus on the Carnatic music tradition of south India, es-
pecially on its melodic characteristics. We overview the
theoretical aspects that are relevant for music information
research and discuss the scarce computational approaches
developed so far. We put emphasis on the limitations of the
current methodologies and we present open issues that have
not yet been addressed and that we believe are important to
be worked on.
1. INTRODUCTION
Though all music traditions share common characteristics,
each one can be recognized by particular features that need
to be identified and preserved. The information technolo-
gies used for music processing have typically targeted the
western music traditions, and current research is emphasiz-
ing this bias even more. However, to develop technologies
that can deal with the richness of our world’s music, we
need to study and exploit the unique aspects of other mu-
sical cultures. By looking at the problems emerging from
various musical cultures we will not only help those specific
cultures, but we will open up our computational methodolo-
gies, making them much more versatile. In turn, we will
help preserve the diversity of our world’s culture [26].
Permission to make digital or hard copies of all or part of this work for
personal or classroom use is granted without fee provided that copies are
not made or distributed for profit or commercial advantage and that copies
bear this notice and the full citation on the first page.
c
2011 International Society for Music Information Retrieval.
The two classical music traditions of the Indian subconti-
nent, Hindustani 1 and Carnatic 2 , are among the oldest mu-
sic and most unique traditions still alive. There are excellent
musicological and cultural studies about them, they main-
tain performance practice traditions and they exist within
real social contexts. Thus, they are an excellent ground on
which to build new information models and a way to chal-
lenge the dominant western-centred paradigms. In this arti-
cle we focus on Carnatic music, the tradition of south-India.
Carnatic music shares with the Hindustani tradition some
basic foundations, such as the basic elements of shruti (the
relative musical pitch), swara (the musical sound of a sin-
gle note), raaga (the melodic mode), and taala (the rhythmic
pattern). Although improvisation plays an important role,
Carnatic music is mainly sung through compositions, dif-
ferently from Hindustani music where improvisation is fun-
damental. Carnatic music is usually performed by a small
ensemble of musicians, consisting of a principal performer
(usually a vocalist), a melodic accompaniment (usually a
violin), a rhythm accompaniment (usually a mridangam),
and a tambura, which acts as a drone throughout the per-
formance. Other typical instruments used in Carnatic per-
formances may include the ghatam, kanjira, morsing, veena
and flute.
The computational study of Carnatic music offers a num-
ber of problems that require new research approaches. Its
instruments emphasize sonic characteristics that are quite
particular and not well understood yet. The concepts of
raaga and taala are completely different to the western con-
cepts used to describe melody and rhythm. Carnatic mu-
sic scores serve a different purpose to those of western mu-
sic. The tight musical and sonic relationship between the
singing voice, the other melodic instruments and the percus-
sion accompaniment within a song, requires going beyond
the modular approaches commonly used in music informa-
tion research (MIR). The special and participatory commu-
nication established between performers and audience in con-
certs, offers great opportunities to study issues of social cog-
1 http://en.wikipedia.org/wiki/Hindustani classical music
2 http://en.wikipedia.org/wiki/Carnatic music

nition. Its devotional aim is fundamental to understand the
music. The study of the song lyrics is also essential to under-
stand the rhythmic, melodic and timbre aspects of Carnatic
music. And many more interesting music aspects could be
identified of relevance to music information processing.
In the next section we focus on the melodic aspects of
Carnatic music, over-viewing the theoretical aspects that are
relevant for MIR and discussing the scarce computational
approaches that have been presented. In the last section we
present open issues that have not yet been addressed and that
we believe are important to be worked on.
2. COMPUTATIONAL APPROACHES TO MELODY
The most fundamental melodic concept in Indian classical
music is raaga. Matanga is the first known person to define
what a raaga is [28]: “In the opinion of the wise, that par-
ticularity of notes and melodic movements, or that distinc-
tion of melodic sound by which one is delighted, is raaga”.
Therefore, the raaga is neither a tune nor a scale [18]. It is
a set of rules which can together be called a melodic frame-
work. The notion that a raaga is not just a sequence of notes
is important in understanding it and for developing compu-
tational models. Also the concept of raga has been chang-
ing with time. Nowadays a given raaga can be described
by properties such as: a set of notes (swaras), their progres-
sions (arohana/avarohana), the way they are intonated using
various movements (gamakaas), and their relative position,
strength and duration (types of swaras). In order to identify
raagas computationally, swara intonation, scale, note pro-
gressions and characteristic phrases are used (Secs. 2.1 and
2.2). Unexploited properties of a raaga include gamakaas
and the various roles the swaras play (Sec. 2.3).
2.1 Swaras and shrutis
In Indian music, swaras are the seven notes in the scale, de-
noted by Sa, Ri, Ga, Ma, Pa, Da and Ni 3 [27]. Except for
the tonic and the fifth, all the other swaras have two varia-
tions each, which account for 12 notes in an octave, called
swarasthanas. There are three kinds of scales that one gener-
ally encounters in Carnatic and Hindustani music theory: a
12-note scale, a 16-note scale and the scale which claims 22
shrutis 4 . The 16-note scale is the same as the 12-note scale
except that 4 of the 12 notes have two names each, in order
to be backward compatible with an older nomenclature.
Few musicians and scholars claim that there are more
shrutis in practice than those explained above. Though many
of them argue the total number to be 22, that itself is de-
bated [9]. A more important question to be asked is whether
they are used in current practice at all. Some musicologists
say that they are no more used [21]. It is also said that
3 This notation is analogous to e.g. Do, Re, Mi, Fa, So, La and Ti.
4 Shruti is the least perceptible interval as defined in Natyasastra [22].
they are wrongly attributed to Bharata, who used shruti to
mean “the interval between two notes such that the differ-
ence between them is perceptible”. Krishnaswamy [13] ar-
gues that the microtonal intervals observed in Carnatic mu-
sic are the perceptual phenomena caused by the gamakaas,
i.e. that these microtonal intervals are what few scholars and
musicians claim as 22 shrutis. However, we believe that
these claims need to be verified with perceptual and be-
havioural studies. In our encounters with most musicians,
we can only conclude that they are unaware of the usage of
22 shrutis in practice. Few musicians who claim they are
used, are not ready to demonstrate them in a raaga. In gen-
eral, more empirical, quantitative and large-scale evidence
needs to be gathered. Our preliminary research on this line
shows no support for the usage of 22 shrutis [25].
The tuning itself, whether it is just-intonation or equi-
tempered, is an issue of debate 5 [12, 25]. Since Indian
classical music is an orally transmitted tradition, perception
plays a vital role. For instance, tuning seldom involves an
external tool. And even the tambura, which is used as a
drone, and thus as a reference for tuning, has a very unsta-
ble frequency. Hence the analysis of empirical data coupled
with perceptual studies are important. In [25] we have car-
ried out an empirical analysis of the stable tunings employed
by some Carnatic and Hindustani singers. The results sug-
gest a clear tendency towards just-intonation in the case of
Carnatic music while, at the same time, they point out to a
strong influence of equi-tempered tuning in the case of Hin-
dustani music.
Fixed tunings are not the whole story. In fact, it is a well
accepted notion that a note (swarasthana) is a region rather
than a point [7,27]. Thus, a fixed, stable tuning for each note
is not as important as it is in, say, western classical music.
In addition, Sa, the tonic, can be any frequency. It depends
on the comfort of the singer or the choice of the instrument
player. A given note can have several variations in intona-
tion depending on the raaga. This variability in intonation
arises from vocal articulations or the pulling of instrument
strings. Even if two raagas have the same scale, the intona-
tion of notes vary significantly. Belle et al [2] have used this
clue to differentiate raagas that share the same scale. They
evaluated their system on 10 audio excerpts accounting for
2 distinct scale groups (two raagas each). They showed that
the use of swara intonation features improved the accuracies
achieved with pitch-class distributions (c.f. [3]). This clearly
indicates that intonation differences are significant to under-
standing and modeling raagas computationally. Levy [16]
analyses the intonation in Hindustani raaga performances
and notes that it is highly variable, and that it does not seem
to agree with any standard tuning system. Subramanian [33]
reports much the same for Carnatic music. These studies
call for the need to understand the extent to which a given
5 http://cnx.org/content/m12459/1.11

Raaga Singer Tested Correctly
identified
Sankarabharanam Nithyasree 5 4
Subbulakshmi 3 2
Balamurali 2 1
Kanakangi Nithyasree 8 6
Ilayaraja 2 1
Karaharapriya Nithyasree 10 6
Table 1. Results of Rajeswari & Geeta’s raaga identification
method.
note can be intonated. In particular, this could be of interest
to differentiate artists and styles.
All these works indicate that a complete characteriza-
tion of swarasthanas must go beyond static frequency mea-
surements and that their dynamics need to be considered.
The problem implies much more than trying to discriminate
whether swarasthanas are tuned to just-intonation, equi-tem-
pered or following 22 shrutis. Much empirical data like the
one reported in [33] and [16] needs to be gathered to investi-
gate the intervals, the range of intonations and the temporal
evolution of each swarasthana.
2.2 Arohana and avarohana
Typically, a raaga is represented using ascending (arohana)
and descending (avarohana) progressions of notes. There
are certain note transition rules that are necessary to be fol-
lowed when performing a raaga. The set of unique notes
in these progressions form a scale. For raaga identification,
Rajeswari et al [31] estimate the scale from the given tune
by comparing it with template scales. Their test data con-
sists of 30 tunes in 3 raagas sung by 4 artists. They use
the harmonic product spectrum algorithm [15] to extract the
pitch, giving the tonic manually. The other frequencies in
the scale are marked down based on the respective ratio with
the tonic. The results obtained are shown in Table 1, which
depicts a 67% accuracy. The authors claim that such a low
accuracy could be due to discrepancies in the manually fed
tonic. But considering that their system identifies only the
swaras that are used in a raaga and no other relevant data, the
result shows that the swaras alone can be very useful. How-
ever, there are raagas which have the same swaras (since the
scales of the raagas they considered are different, this is not
an issue in their study).
Shetty et al [29] use a similar approach when they try
to recognize raagas. The features extracted are the individ-
ual swaras and their relation in arohana-avarohana (swara
pairs). The features are represented as bit sequences which
are later converted to decimal values. These features are
used for training a neural network. They report an accuracy
of 95% over 90 tunes from 50 raagas, using 60 tunes as train-
ing data and the remaining 30 tunes as test data. However,
such a high accuracy is questionable due to the few data per
class used. Moreover, no cross-fold validation was done.
Sahasrabudde et al [23] model the raaga as finite automa-
ta. A finite automata has a set of states between which the
transitions take place. In the case of raaga, the swarasthanas
are the states and the note transitions are observed. This idea
is used to generate a number of audio samples for a raaga,
which they claim are technically correct and indistinguish-
able from human compositions. Inspired by this, Pandey et
al [17] use HMM models to recognize the raagas. The rules
to form a melodic sequence for a given raaga are well de-
fined in the musicology literature [24] and the number of
notes is finite. Therefore, intuitively, HMM models should
be good at capturing those rules in note transitions imposed
by arohana and avarohana patterns (at least the first-order,
simpler ones).
Each raaga has also a few characteristic phrases. They
are called swara sancharas in Carnatic and pakads in Hin-
dustani. These phrases are said to be very crucial for con-
veying the feeling of the raaga [9]. Typically, in a concert,
the artist starts by singing these phrases. They are the main
clues for the listeners to identify which raaga it is. Pandey et
al have complemented their approach with values obtained
from two modules that match characteristic phrases, taking
advantage of this information. In one such module, char-
acteristic phrases are identified with a substring matching
algorithm. In the other one, they are identified by counting
the occurrences of frequency n-grams in the phrase.
The other important contributions by Pandey et al in-
clude two heuristics to improve the transcription of Indian
classical music: the hill peak heuristic and the note dura-
tion heuristic. As mentioned, Indian music has a lot of mi-
cro tonal variations which makes even the monophonic note
transcription a challenging problem [17]. The two heuristics
proposed in their approach try to get through these micro
tonal fluctuations in attaining a better transcription. The hill
peak heuristic states that a significant change in the slope of
a pitch contour (or the sign reversal of such slope) is closely
associated with the presence of a note. The note duration
heuristic considers only the notes that are played for at least
a certain span of time. The approach was tested on two raa-
gas. Table 2 shows the results obtained by using HMMs
alone, and by complementing the models with characteristic
phrase matching. Not much can be said about the reliability
of the features they used since the number of classes con-
sidered were just two. But the advantage of characteristic
phrase matching is evident.
Sinith et al [30] also used HMMs of raagas to search for
musical patterns in a catalogue of monophonic Carnatic mu-
sic. They build models for 6 typical music patterns corre-
sponding to 6 raagas (they report a 100% accuracy in iden-

Raaga Samples HMM HMM +
Phrase matching
Yaman Kalyan 15 80% 80%
Bhupali 16 75% 94%
Total 31 77% 87%
Table 2. Accuracy of raaga identification reported in [17].
tifying an unknown number of tunes into 6 raagas). HMMs
are also used by Das and Choudary [6] to automatically gen-
erate Hindustani classical music.
Chordia and Rae [3] use pitch class profiles and bi-grams
of pitches to classify raagas. The dataset used in their sys-
tem consists of 72 minutes of monophonic instrumental (sa-
rod) data in 17 raagas played by a single artist. Again, the
harmonic product spectrum algorithm [15] is used to extract
the pitch. Note onsets are detected by observing the sudden
changes in the phase and the amplitude of the signal. Then,
the pitch-class profiles and the bi-grams are calculated. It
is shown that bi-grams are useful in discriminating the raa-
gas with the same scale. They use several classifiers com-
bined with dimensionality reduction techniques. The feature
vector size is reduced from 144 (bi-grams) + 12 (pitch pro-
file) to 50 with principal-component analysis. Using just the
pitch class profiles, the system achieves an accuracy of 75%.
Using only bi-grams of pitches, the accuracy is 82%. Best
accuracy of 94% is achieved using a maximum a posteriori
rule with a multi-variate likelihood model. Comparison to
other classifiers is shown in [3].
2.3 Unexploited properties of raaga
2.3.1 Gamakaas
In Carnatic music the various forms of pitch movements
are together called gamakaas. A sliding movement from
one note to another or a vibrato are examples of gamakaas.
There are various ways to group these movements, but the
most accepted classification speaks of 15 types of gamakaas.
Gamakaas are not just decorative items or embellishments,
but very essential constituents of a raaga [9]. Each raaga
has some characteristic gamakaas. Thus, the detection of
gamakaas is a crucial step to model and identify raagas.
A gamakaa is often represented using discrete notes, but
it does not necessarily mean that one plays them using dis-
crete steps. The representation is only a handy expression
of a more continuous sounding pattern, which is difficult
to represent on the paper. A gamakaa is almost always a
smooth change in the dynamics of a pitch contour. Similar
concepts are used to describe the pitch inflections in Hin-
dustani music [19]. Owing to their tremendous influence on
how a tune sounds, the gamakaas and the related pitch in-
flections in Hindustani music are often considered the soul
of Indian classical music.
There are two major issues that make identifying a gama-
kaa a challenging problem. First, it requires a very precise
pitch transcription. Second, the variations found for differ-
ent artists in performing a gamakaa complicate it further.
Krishnaswamy [14] and Subramanian [33] report such vari-
ations across different artists performing the same gamakaa.
They also propose some theoretical guidelines to resolve the
second problem to some extent. These variations should
be exploited in performers’ computational modeling, a field
that lacks much research in the case of Indian classical mu-
sic.
2.3.2 Various roles played by the notes
In a given raaga, not all the notes play the same role. Though
two given raagas have the same set of constituent notes, their
functionality can be very different, leading to a different
feeling altogether [34]. For example, some swaras occur
frequently, some are prolonged, some occur either at the be-
ginning or the end of the phrases, etc. In addition, there are
alankaras, patterns of note sequences which are supposed to
beautify and instil feelings when listened to.
Though emotion is a subjective issue, it gets into almost
every discussion involving raagas. That is because each
raaga is said to evoke characteristic emotions. To test this
hypothesis, Chordia and Rae [4] have conducted a survey
to check whether Hindustani raagas elicit emotions consis-
tently across listeners. Positive results are reported, jointly
with the musical properties like relative weight of the notes,
which partially explain the phenomenon. Koduri et al [11]
have conducted a similar survey with Carnatic raagas. Though
not as significant as the pattern reported by Chordia et al, the
results indicate that Carnatic raagas elicit emotions which
are consistent across listeners. Wieczorkowska et al [35]
tests if raagas elicit emotions, and also arrive at a mapping
between melodic sequences of 3 or 4 notes and the elicited
emotions. Their work suggests that different compositions
in the same raaga might elicit different emotions, what is
consistent with the observations made by Koduri et al [11].
Wieczorkowska et al note that these melodic sequences are
related vaguely to the subjects’ emotional responses. Anoth-
er interesting observation is the significance in the similar-
ity between the responses of people from various cultures,
which is consistent with the observations made in a previous
study conducted by Balkwill et al [1].
3. OPEN ISSUES: GAMAKAAS, TAALAS,
INSTRUMENTS AND IMPROVISATION
Little research has been carried out on Carnatic music and
even less on the specific characteristics that makes it so spe-
cial. Few proposed computational approaches have focused
on raaga recognition and the results are quite preliminary

given that the data used is not representative of the exist-
ing variety of raagas. The high accuracies reported might
be due to the limited number of raagas used and the small
sizes of the datasets. Moreover, important properties of the
raagas, like their specific use of gamakaas, have not been
exploited yet, and issues beyond recognition have neither
been approached. We hypothesize that, as more represen-
tative datasets are gathered, the features used will not be
sufficient to discriminate the raaga classes. Features such as
pitch-class profiles and pitch-class dyad distributions infer
partial information about the raagas. But the other roles of
notes are not evident, which need to be exploited. Symbolic
scores can also be used for building more complex models,
especially to model the characteristic melodic movements
of particular raagas.
While raaga is the fundamental concept related to melody,
taala is the fundamental concept related to rhythm [34]. A
taala is a rhythmic cycle, which is divided into specific un-
even sections, each of them subdivided into even measures.
The first beat of each taala section is accented, with notable
melodic and percussive events. The characteristics of a taala
are related to the main instrument used to emphasize the
rhythmic aspect in a song, the mridangam. Understanding
the acoustics of the mridangam and how it is played, is fun-
damental to model the taalas. Sambamoorty [24] lists all
taalas and provides the description for each. The recog-
nition of the different types of strokes to play the mridan-
gam, bols, is an open topic. Current MIR research on drum
transcription uses small numbers of drum stroke classes and
each class is associated with a specific (single) drum, usu-
ally based on the typical western drum set. With mridan-
gam, multiple bols are associated to each drum, and given
that is a tuned instrument, the recognition of the bols have to
take into account both timbre and pitch information. Some
work has been done on the recognition of bols in Hindus-
tani music, with the tabla [8] [5], but no research has been
carried out in Carnatic music, with the mridangam. There
is also no research focusing on the recognition or classifi-
cation of taalas. As the musician always tries to embellish
the taala, there is a strong variation from performance to
performance, and the rhythmic complexity obtained is enor-
mous. The main goal would be to gain insensitivity to these
variations in order to classify taalas or, otherwise, to model
these variations for understanding performance and impro-
visation. For this research we need to use top-down or other
contextual information to make sense of the audio data, for
example there is a well-defined structure to improvisation
which should be exploited [9].
We have reported on previous work that has verified whe-
ther raagas elicit emotions and tried to map the musical fea-
tures which are responsible for such phenomenon. Besides
the note sequences, another important aspect of Indian clas-
sical music which could play a crucial role in eliciting emo-
tions is gamakaa. However, there are no studies which re-
port their effect so far. The kind of instruments used and the
rhythmic aspects also need to be accounted when dealing
with emotional aspects.
At the level of musical instruments there is practically
nothing done. Physical modeling of their many non-linear
behaviours is quite complex and the lack of instrument stan-
dardization does not help. Some research has been done on
modeling north-Indian instruments like the tabla and sitar
[10] and there have been a few attempts in developing sound
synthesis systems [32]. The timbre of the tambura is at
the basis of the Indian sound. It has a special overtone-
rich sound, a sustained ”buzzing” resulting from the wide
and arched bridge on which the strings rests and of the cot-
ton thread placed between the strings and the bridge. This
type of string termination results in a quite complex acoustic
system first discussed by Nobel Prize winning physicists C
V Raman [20] and for which current F0-detection methods
perform very poorly.
The performance practice tradition has not been studied
at all. Music performance is mainly learned by imitation,
without much use of symbolic representations. The vari-
ability in performances of the same song is quite large, es-
pecially due to the importance of improvisation. The same
composition sung by two artists can be different in many
musical and expressive facets. These differences may chal-
lenge the version identification methods developed for west-
ern commercial music. In addition to the compositional
forms, there are many improvisatory forms that are perfor-
med with well-defined structural criteria [9].
Through the article we have mentioned a number of char-
acteristics of Carnatic music that deserve to be studied. Gi-
ven that this music tradition is so different from the ones
used to develop the current computational methodologies,
there is a need to deal with some more fundamental issues
related to music information processing. We need to study
how the musical concepts and terms in Indian music are un-
derstood, specifying proper ontologies with which to frame
our work. Also the cultural and community aspects of the
music are so important that, without studying them, we will
not be able to develop proper musical models. In summary,
to approach the computational modeling of Carnatic music,
making justice to its richness, it is fundamental to take a
cultural approach and, thus, take into account musicological
and contextual information.
4. ACKNOWLEDGEMENTS
The research leading to these results has received funding
from the European Research Council under the European
Union’s Seventh Framework Programme (FP7/2007-2013)
/ ERC grant agreement 267583 (CompMusic).

5. REFERENCES
[1] L.L. Balkwill and W.F. Thompson. A cross-cultural investiga-
tion of the perception of emotion in music: Psychophysical and
cultural cues. Music Perception, 17(1):43–64, 1999.
[2] S. Belle, R. Joshi, and P. Rao. Raga identification by using
swara intonation. Journal of ITC Sangeet Research Academy,
23, 2009.
[3] P. Chordia and A. Rae. Raag recognition using pitch-class and
pitch-class dyad distributions. In Proc. of ISMIR, pages 431–
436, 2007.
[4] P. Chordia and A. Rae. Understanding emotion in raag: An
empirical study of listener responses. In Proc. of International
Symposium on Computer Music Modeling and Retrieval, pages
110–124, 2009.
[5] P. Chordia and A. Rae. Tabla gyan: an artificial tabla Impro-
viser. In Proc. of International Conference on Computational
Creativity, 2010.
[6] D. Das and M. Choudhury. Finite state models for generation
of Hindustani classical music. In Proc. of International Sym-
posium on Frontiers of Research in Speech and Music, 2005.
[7] A.K. Datta, R. Sengupta, N. Dey, and D Nag. Experimental
analysis of shrutis from performances in Hindustani music.
ITC Sangeet Research Academy, 2006.
[8] O. Gillet and G. Richard. Automatic labelling of tabla signals.
In Proc. of ISMIR, 2003.
[9] S. R. Janakiraman. Essentials of musicology in south Indian
music. The Indian Music Publishing House, 2008.
[10] A. Kapur, P. Davidson, P. Cook, W. Schloss, and P. Driessen.
Preservation and extension of traditional techniques: digitiz-
ing north indian performance. Journal of New Music Research,
34(3):227–236, 2005.
[11] G. K. Koduri and B Indurkhya. A Behavioral Study of Emo-
tions in South Indian Classical Music and its Implications in
Music Recommendation Systems. In SAPMIA, ACM Multime-
dia, pages 55–60, 2010.
[12] A. Krishnaswamy. On the twelve basic intervals in south indian
classical music. In Proc. of Audio Engineering Society Conven-
tion, 2003.
[13] A. Krishnaswamy. Inflexions and microtonality in south Indian
classical music. In Proc. of International Symposium on Fron-
tiers of Research in Speech and Music, 2004.
[14] A. Krishnaswamy. Multi-dimensional musical atoms in South-
Indian classical music. In Proc. of International Conference on
Music Perception and Cognition, 2004.
[15] K Lee. Automatic chord recognition from audio using en-
hanced pitch class profile. In Proc. of the International Com-
puter Music Conference, 2006.
[16] M Levy and N. A. Jairazbhoy. Intonation in North Indian
Music: A Select Comparison of Theories with Contemporary
Practice. Aditya Prakashan, New Delhi, 1982.
[17] G. Pandey, C. Mishra, and P. Ipe. Tansen: A system for auto-
matic raga identification. In Proc. of Indian International Con-
ference on Artificial Intelligence, pages 1350–1363, 2003.
[18] H. S. Powers. The Background of the south Indian raaga-
system. PhD thesis, Princeton University, 1959.
[19] Pratyush. Analysis and classification of ornaments in north In-
dian (Hindustani) classical music. Master’s thesis, Universitat
Pompeu Fabra, 2010.
[20] C. V. Raman. On some Indian stringed instruments. In Proc.
Indian Assoc. Cultiv. Sci., volume 33, pages 29–33, 1921.
[21] N Ramanathan. Shrutis according to ancient texts. Journal of
the Indian Musicological Society, 12(3):31–37, 1981.
[22] A Rangacharya. The Natyasastra. Munshiram Manoharlal
Publishers, 2010.
[23] H.V. Sahasrabuddhe and R. Upadhye. On the computational
model of raag music of india. In Workshop on AI and Music:
European Conference on AI, 1992.
[24] P. Sambamoorthy. South Indian Music. The Indian Music Pub-
lishing House, 1998.
[25] J. Serra`, G. K. Koduri, M. Miron, and X. Serra. Assessing the
tuning of sung Indian classical music. In Proc. of ISMIR, 2011.
[26] X. Serra. A multicultural approach to music information re-
search. In Proc. of ISMIR, 2011.
[27] V. Shankar. The art and science of Carnatic music. Music
Academy Madras, Chennai, 1983.
[28] P.L Sharma and K Vatsayan. Brihaddeshi of Sri Matanga Muni.
South Asian Books, 1992.
[29] S. Shetty and K. K. Achary. Raga mining of Indian music by
extracting arohana-avarohana pattern. International Journal of
Recent Trends in Engineering, 1(1), 2009.
[30] M. S. Sinith and K. Rajeev. Hidden markov model based recog-
nition of musical pattern in south Indian classical music. In
Proc. IEEE International Conference on Signal and Image
Processing, 2006.
[31] R. Sridhar and T.V. Geetha. Raga identification of carnatic mu-
sic for music information retrieval. International Journal of
Recent trends in Engineering, 1(1):1–4, 2009.
[32] M Subramanian. Synthesizing Carnatic music with a computer.
Journal of Sangeet Natak Akademi, pages 16–24, 1999.
[33] M Subramanian. Carnatic ragam thodi - pitch analysis of notes
and gamakams. Journal of the Sangeet Natak Akademi, pages
3–28, 2007.
[34] T. Viswanathan and M.H. Allen. Music in south India. Oxford
University Press, 2004.
[35] A Wieczorkowska, A. Datta, R Sengupta, N Dey, and
B Mukherjee. On search for emotion in Hindusthani vocal mu-
sic. Advances in Music Information Retrieval, pages 285–304,
2010.