iSargam: music notation representation for Indian Carnatic music
© Mammen et al. 2016
Received: 15 May 2015
Accepted: 5 February 2016
Published: 16 February 2016
Indian classical music, including its two varieties, Carnatic and Hindustani music, has a rich music tradition and enjoys a wide audience from various parts of the world. The Carnatic music which is more popular in South India still continues to be uninfluenced by other music traditions and is one of the purest forms of Indian music. Like other music traditions, Carnatic music also has developed its musicography, out of which, a notation system called Sargam is most commonly practiced. This paper deals with development of a music representation or encoding system for the Sargam notation scheme which enables easy music notation storage, publishing, and retrieval using computers. This work follows a novel idea of developing a Unicode-based encoding logic and allows storage and easy retrieval of music notation files in a computer. As opposed to many existing music representation systems for western music notation, iSargam is the only music notation encoding system developed for Indian music notation.
The textual descriptions of music notations can be further classified as record-based, command-based, symbolic codes, and LISP-based. While systems like DARMS , Guido  etc. used symbolic codes, systems like CMN  used command-based representations. A few examples of LISP-based representations are MUZACS , Rhythm-Editor of Patch Work , and CMN . Also, to take advantage of XML features like structuring and portability, many popular XML-based notations like MusicXML , MEI  , and WEDELMUSIC  emerged.
2 Background and history
Indian classical music is one of the oldest music traditions in the world, and it enjoys the next position to western music in its popularity. The music tradition of India can be divided into two large traditions, namely, Hindustani and Carnatic music. The Hindustani classical music which is predominant in the northern part of the Indian subcontinent, originates from the ancient Vedic, Persian, and many folk traditions. While the Carnatic classical music, uninfluenced by non-Indian music traditions, is the purest form of Indian music and is prevalent in the southern parts of the Indian subcontinent. It is generally homophonic in nature with emphasis on vocal music. If performed on an instrument, it assumes a singing style.
Carnatic music is usually performed by an ensemble of musicians consisting of a principal performer, usually a vocalist, accompanied by a rhythm instrument, melodic instrument, and a monophonic drone instrument.
The tradition of using music notations for singing was practiced from Vedic times. But there was no uniformity in the notation system used from time to time. However, all the notations developed so far can be classified as script as opposed to staff notation used for western music. Different systems of notation were prevalent in each period of history in both Carnatic and Hindustani traditions. Also, many notation systems existed in parallel.
The main objective of this work is to develop a unified representation system for storing Indian Carnatic music notations in computer files. The system aims at encoding music notation symbols and other associated information like title, composer name etc., while ignoring the layout specific details. Most of the encoding systems developed so far are intended at encoding western music staff notation or a derivate of it. This work aims at developing a machine readable music notation system for Carnatic music which can support playback, printing, retrieval, and searching within the composition.
Since Unicode is currently supported by most text editors and web browsers, the iSargam file can be read directly and displayed without the need for any additional code similar to the ASCII-based text files. Additionally, they can employ plug-ins to provide different styles and layouts for printing purposes. Since most of the current programming languages support Unicode, application programmers and music analysts can easily build algorithms for the music information retrieval or computer-based analysis of the iSargam music databases.
The next section gives details about Sargam notation system, and the section following provides details of related works done elsewhere. We then describe the iSargam encoding system, explaining its approach and encoding algorithm. For increasing readability of western readers, we give comparisons with western music concepts wherever applicable. We also present some example encoding compositions as proof of our approach.
3 The Sargam notation system
Sargam notation is a music notation language for Carnatic music. Each notation starts with specification of raga, tala, and mela. Sometimes, the notes used in ascending scale and descending scale, known as arohana and avarohana, are explicitly defined in the start of the composition. This is followed by the time signature and actual music notation following it as illustrated in Fig. 2. In this section, we attempt to describe the terminologies used and the notation scheme.
Raga  is one of the most distinguished features of Carnatic music. The raga can be defined by a melodic scheme characterized by a definite scale or notes, order or sequence in which the notes can be used, melodic features, pauses and stresses, and tonal graces. Some ragas define the same set of music notes (swara) but are still differentiated by some other features like the order of appearance of swara, melodic punctuation, accent, intonation, and melodic phrases. The raga in Carnatic music is analogous to key signatures in western music.
A related field is the specification of arohana and avarohana. The arohana (meaning “ascending”) follows from the raga specification and explicitly lists the set of allowed swara syllables when the music is following an ascending flow. Similarly, the avarohana (meaning “descending”) specifies the name of the parent raga.
The term “tala”  refers to the rhythm system which controls and establishes the music. There are hundreds of defined rhythm styles (talas) in Indian music. The name of the tala used in the notated composition is given above the notation as in Fig. 2.
Basic elements of Indian Carnatic music rhythms (tala)
Duration in aksharakala units
|3, |4, |5,|6 |7, |9
The rhythm pattern is repeated in a cyclic manner throughout the music and hence is known as “avarta,” which means repetition. The basic rhythm used is notated at the beginning of the notation, and notes are grouped according to it as seen in Fig. 2. The grouping method used is similar to grouping notes according to time signature as in western music notation but with bars of different measure.
Mela or Melakartas  are parent ragas from which the other ragas evolved. There are twenty-two of them. A distinguished feature of the mela or parent ragas is that they contain all the seven notes in order. The melakartas have a numbering scheme and are identified by the number. The parent raga of the composition is so indicated by an integer number as illustrated in Fig. 2.
The notation style is primarily script-based where the note symbols are placed on a straight line. Suitable signs and symbols are also used to indicate various other musical features. A detailed review of the Sargam transliteration scheme can be found in .
As shown in Fig. 2, the notation part starts with tala symbols called “anga” which group the other music notes according to time duration. The symbols used in notation can be classified as music note (swara), gamaka symbols, and other symbols. The following section details the concept and notation of musical note, gamaka, and other symbols.
3.4.1 Musical note
A swara or music note usually denotes the note name indicating the pitch, duration, octave, and whether it is played with expressions (called gamaka) or not. The notes are named differently according to its pitch as shadja (sa), rishabha (ri), gandhara (ga), madhyama (ma), panchama (pa), dhaivata (dha), and nishada (ni) and is abbreviated as given in brackets. The current style of written forms is with vowels removed and expressed with single letters as shadja (S or s), rishabha (R or r), gandhara (G or g), madhyama (M or m), panchama (P or p), dhaivata (D or d), and nishada (N or n).
3.4.2 Octave or sthayi
The duration of each music note is measured in terms of aksharakala, which is the unit of measurement in the Indian tala (rhythm) system and is analogous to “beats” in western music. To denote the duration of a music note, uppercase or lowercase letters with or without comma, semicolon, or with underline or over line are used. A swara letter in lowercase indicates one aksharakala duration, and an upper case swara letter indicates two aksharakala duration. A comma placed near a music note increases its duration by one aksharakala and a semicolon by two. Similarly, a single horizontal line over the swara reduces the swara duration to its half and double over or under line reduces it to its quarter.
The duration of a rest note is indicated using the necessary number of semicolon or comma symbols placed inside simple parenthesis, e.g., (,;).
3.4.4 Note variety
In Carnatic music, each music note (swara) can represent more than one pitch value, usually two, according to the raga followed by the composition. Because of this characteristic, they are called “note varieties” as each swara note provides many colors to choose from.
Generally, there are no special signs or symbols to represent the variety of the note. This information is implicitly associated with the raga of the song. However, some subscripted numerals with Swara symbols are rarely used to denote sharp and flat varieties of the notes which comprise the twelve “swarasthanas”  . For example, r1 denotes the musical note “Sudhhari” or “Komalari” and r2 denotes “Chatusrutiri” or “Tivrari”  .
3.4.5 Additional symbols
List of additional symbols used and their notation
Name of symbol
/ and \
Rarely, some notes which are not part of the raga specification are used, and such notes are represented by an asterisk mark over the swara symbol. The repeat symbol, usually found at the end of an avarta (measure) denotes that the portion of music should be repeated. A stressed note, denoted by letter “w” over the swara, is similar to staccato in functionality, as used in western music. The gamaka mark, represented by a tilde symbol over the swara symbol symbolizes ornamentation which is of utmost importance to Indian music. A music phrase, unlike in western music, represents a set of musical notes which has to be sung together in one breath duration and is symbolized by hyphens at the start and end of the phrase.
3.5 Notation arrangement
The music notes with adjoint symbols are written on a straight line similar to tonic solfa notation in western music. The music notes are then grouped according to the rhythm structure (tala) of the composition, which is similar to the grouping of notes with time signatures in western music notation. Here, we explain the grouping mechanism in comparison with grouping in western notation for easy understanding of readers.
4 Related works
Music representation systems encompass musical information in any of the three levels: sound, music notation, or data for analysis . Music notations are generally an encoding of abstract representations of music. They contain instructions for performance and representation of sound. Several structured representation like the hierarchical music structures representation , “Music Structures” , “TTrees” , hierarchical representation of scores , musical events , musical tones , abstract datatype representation in , generative approaches like in grammars , Petri Nets , Markov Models , and object-oriented approaches like SmOKe , Aspect Music , and graph-based approaches , for representing musical data.
Music representation systems can be classified as audio signal representations, resulting from the recording of sound sources or from direct electronic synthesis, and symbolic representations which represent discrete musical events such as notes, rhythm etc. . The proposed system is a symbolic representation, and is content-aware and can relate musical events to formalized concepts of Carnatic music theory.
The musical representation systems can also be classified according to the encoding system (format) used for storing the information. Thus, it can be classified into binary, ASCII-based, XML-based, and proprietary formats. The popular binary formats are MP3, WAV etc. The ASCII formats like Cadenza , DARMS , Guido , ENP , LilyPond , and Humdrum  encode musical score information using ASCII-based text. They can be further classified as record-based, command-based, codes, and LISP-based. The XML formats like MIDI XML , MusicXML , MEI  etc. provide hierarchical XML representation of musical information. Also, many popular score-writing programs like Rhapsody and Sibelius use proprietary formats. We propose to use Unicode standards to encode the Carnatic music notation called “Sargam” (or modern notation), and it can be considered as the first Unicode-based music representation system.
Even though most of these music representation systems were evolved around western music tradition/notation, there were a few attempts to extend its applicability to other regional music traditions like Korean , Greek , **Bhat  etc. Unlike these extensions of western music, the proposed system is a unique approach to representation of South Indian Carnatic music based on Indian music theory.
5 The iSargam language
As opposed to many ASCII-based representations like DARMS , Guido  etc., iSargam is formed as a Unicode-based music notation representation language. That means we use various Unicode symbols to represent musical entities in Carnatic music.
In this section, we describe the iSargam representation system by explaining its approach, encoding logic, and algorithm.
Before describing the encoding logic, we would like to illustrate a few concepts which we developed as part of the encoding logic, viz, singleton/grouped entity, and music constituent.
5.1.1 Singleton/grouped entity
The musical symbols used in Sargam notation are classified as singleton or grouped entity according to whether they have meaning or sense in single form or they make sense only when they combine with another musical entity. For example, anumandra, the octave specification symbol makes sense only when it is joined with a musical note (Swara). Singleton musical entities are always found independent in the notation and have semantics of their own. An example of singleton entities are tala marking symbols, anga/avarta mark etc. This classification among music symbols is required due to difference in encoding single and group entities, where group entity symbols can be encoded together only and not individually.
5.1.2 Music constituent
It may be noted that the swara syllable alone is a complete musical constituent since it already contains octave and duration information.
The basic element can be grouped according to some rhythmic pattern or it can be further augmented with additional symbols or other music notes, forming various grouped entities. In our approach, the former is called rhythmic group and the latter is called notational grouping. This latter is again classified into intra notational and inter notational group entities. Inter notational groupings are always associated with music constituents.
Intra notational grouping occurs when the music constituent is further augmented by adding parameters which apply in a single note level. This is denoted by adding extra signs or symbols to the base syllable. The musical entities in this category are stressed note symbol (w), gamaka mark (~), foreign note symbol (*), violin marks, upward stroke of the bow (v), and downward stroke of the bow (^). In case of inter notational grouping, multiple musical notes are grouped together, mostly to give a musical expression such as a musical phrase and ascending or descending glides.
5.2 Encoding logic
Having defined the basic terminologies, now we attempt to present our encoding logic. Initially, our system maps every Sargam notation symbol to a Unicode symbol. The chosen Unicode character resembles the Sargam notation symbol used. Each symbol is also assigned a priority number.
The iSargam system chooses the unique numbers carefully so as to make sure that the corresponding Unicode character almost fully resembles the actual music notation in appearance, even in the case of the grouping or joining of music notation symbols. The advantage here is that Unicode symbols appear discrete in encoding, which favors easy identification of music entities for music processing, but in appearance, it appears joined, resembling the original notation. Also, it may be noted that in such a representation, a combined notation can be easily split to its constituent basic music entities.
We use Unicode full width forms for standalone music elements like swara syllable or duration, and Unicode combining diacritical marks for adjunct symbols like octave, stress, foreign note indication, duration symbols which symbolize duration less than one unit, etc. More specifically, all the intra notational symbols and violin marks are represented by combining diacritical marks. Additionally, we use Unicode full width symbols for representing symbols in the rhythmic group like anga, avarta symbols, which are analogous to measure and bar markings for western music, and other rhythm-specific elements like laghu, plutum etc., which mark the number of beats within a measure.
The encoded file consists of various sections, viz, the header, rhythm markup, and actual composition, explained as follows.
5.2.1 Header section
The header section accompanies every music notation. It mainly consists of two sections, viz, a compulsory part and an optional part. The compulsory part is known as the music description part, and it specifies the most important elements for interpreting the notation. These most important fields are raga name and tala name. The optional part consists of the fields composition title, composer, arohana/avarohana, and mela. The header elements are considered as keywords which are case-insensitive and are separated by a colon character. These values are case-sensitive.
5.2.2 Tala (rhythm) markup section
The tala section marks the rhythm pattern of the composition. Usually, a rhythm pattern is defined as a combination of its basic elements called angas, as described in the previous section. For encoding, each anga symbol is assigned a Unicode-based identifier and is separated from the others by using the vertical line symbol (U+007C). The avarta is marked by a double pipeline symbol (U+01C1) at the beginning and the end as shown in Table 4.
5.2.3 Music notation section
This section contains the notation of the actual composition. It consists of music constituents along with required signs and symbols with notational and rhythmic grouping. The encoding of the actual music notation is illustrated in the following subsections.
Encoding of rhythm grouping
The music notes are grouped according to the tala specification, splitting them into many anga and avarta as described in the previous section. The angas are separated by vertical line symbol (U+007C), and avarta is marked by double pipeline symbol (U+01C1) at the beginning and the end.
Sometimes, a repeat symbol is inserted in front of the avarta end to denote repetition of an avarta. The symbol used is “(r)”and the encoding is encoded by the Unicode symbol parenthesized Latin small letter R (U+24AD).
Encoding of music note
The encoding strategy followed for a pitched music note and an unpitched note is different. The encoding logic for an unpitched note is straightforward like a singleton entity. But a pitched note is regarded as a grouped entity and we use priority-based encoding for this set as illustrated in Fig. 6.
Unlike the encoding strategy used for singleton entities, encoding for grouped entities is done together and not as individual elements. It can be observed that any pitched note is an extension of a pitched music constituent. In case of a pitched music constituent, the swara symbol is followed by octave and duration symbols, as mentioned in the previous section. Additionally, musical notes can contain additional characters which augment the basic music note like stressed note symbol, foreign note symbol etc. The musical note may then be part of another group in case of occurrence of musical phrase or glide expressions. It might look straightforward to assign symbol priorities in the same order. But this does not work due to a difference in the type of notation symbols used. So, iSargam develops a new encoding logic for notating a music note, which is explained here. In this context, we would like to redefine the general syntax of music notes given in the previous section to enable easy encoding.
The <duration-1> symbol consists of duration symbols which have duration of less than one aksharakala unit, and we use Unicode diacritics symbols to represent them. <additional-symbols> are also represented by Unicode diacritics symbols. The <duration-2> consists of duration which have duration greater than one aksharakala unit, and we use basic Latin Unicode symbols comma and semicolon to represent them.
Inter notational grouping
So, the encoding also follows a simple strategy of placing the Unicode of the backward or forward slash between the two musical note elements. Even though they are grouped symbols, the encoding strategy used encodes them in a straightforward manner by just inserting the Unicode symbol solidus (U+002F) for ascending glide and reverse solidus (U+OO5C) for descending glide.
5.3 Encoding algorithm
The proposed work is an encoding system for Carnatic music notation. The limitation of this work is that it only stores musical information in a retrievable form. A possible extension of this work is to integrate search and retrieval mechanisms upon the encoded form.
Most of the popular music retrieval systems employ search mechanisms with note patterns fed into the system as a set of note or swara symbols using the basic ASCII character set. Similarly, most of the query-by-humming-based music information retrieval systems internally convert the user-hummed query into music notations before the actual comparison process. Since iSargam is a Unicode-based encoding system, an extension of the existing MIR systems to support iSargam files or an application of the MIR approaches to the iSargam musical database will only require an integration of a simple ASCII to Unicode conversion module. The same approach can be used to apply the musical analysis mechanisms to the iSargam files or database.
Another suggested improvement is to integrate the work with a knowledgebase developed for Indian music and to incorporate a playback facility with the editor.
This work has been carried out as part of a sponsored project entitled “Representation, Retrieval and Analysis Mechanisms for South Indian Carnatic Music using Computers” with the sponsorship of Ministry of Culture, Government of India, for a period of 2 years. The authors acknowledge the financial support made the agency to carry out the proposed work.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.
- M Balaban, Musical structures: interleaving the temporal and hierarchical aspects to music. (In Understanding Music with AI: Perspectives in Music Cognition, MIT Press, Cambridge, 1992), pp. 110–138Google Scholar
- N Han-Wen, N Jan, LILYPOND, A System for Automated Music Engraving (Paper presented at the XIV Colloquium on Musical Informatics, Firenze, 2003)Google Scholar
- D Huron, Music information processing using the Humdrum Toolkit: concepts, examples, and lessons. Comp. Music J. MIT Press 26(2), 11–26 (2002)View ArticleGoogle Scholar
- HH Hoos, KA Hamel, K Renz, J Kilian, Representing score-level music using the GUIDO music notation format. (Computing in Musicology, MIT Press, Cambridge, 2001), p. 12Google Scholar
- HS Field-Richards, Cadenza: a music description language. Comput. Music. J. 17(4), 60–72 (1993)View ArticleGoogle Scholar
- Solesmes, Antiphonale Monasticum, vol 1, (Paraclete Press, 2005), pp. 542–543Google Scholar
- G Bays, ScoreSVG: a new software framework for capturing the semantic meaning and graphical representation of musical Scores Using Java2D, XML, and SVG (Master's Thesis, College of Arts and Sciences, Georgia State University, 2005)Google Scholar
- MM Erin, R Jenny, Saffran, music and language: a developmental comparison. Music. Percept. 21(3), 289–311 (2004)View ArticleGoogle Scholar
- Center for Computer Research in Music and Acoustics, Common Music Notation, (Stanford University, 2010), https://ccrma.stanford.edu/software/cmn/cmn/cmn.html. Accessed on 12 April 2015
- W Kornfeld, Machine Tongues VII: LISP. Comput. Music. J. 4(22), 6–12 (1980)View ArticleGoogle Scholar
- M Laurson, Dissertation, Sibelius Academy, 1996Google Scholar
- M Good, G Actor, Using MusicXML for file interchange, In Proceedings Third International Conference on WEB Delivering of Music, Leeds, UK, September 15–17, 2003 (IEEE Press, Los Alamitos, 2003), p. 153Google Scholar
- P Roland, Design Patterns in XML Music Representation (Paper presented at the Fourth International Conference on Music Information Retrieval, University of Johns Hopkins, Baltimore, 2003)Google Scholar
- R Perry, The music encoding initiative, in Proceedings First International Conference on Musical Application using XML, ed. by H Goffredo, L Maurizio, 2002Google Scholar
- P Bellini, P Nesi, WEDELMUSIC Format: An XML Music Notation Format for Emerging Applications, First International Conference on WEB Delivering of Music (IEEE Computer Society, Washington, 2001), pp. 79–87Google Scholar
- PP Narayanaswami, VS Jayaraman (eds.), Sangita Sampradaya Pradarsini English Web Edition, 2006. http://ibiblio.org/guruguha/ssp.htm. Accessed on 15 June 2014Google Scholar
- P Sambamurthy, A Practical Course in Karnatik Music, 1st edn. (The Indian Music Publishing House, India, 1963), pp. 23–56Google Scholar
- L Issac, Theory of Indian music. (Shyam Printers, 1967), pp. 18–24Google Scholar
- S Bhagyalekshmy, Ragas in Carnatic Music, (CBH Publications, 2003)Google Scholar
- P Hill, S Holland, R Laney, An introduction to aspect oriented music representation. Comput. Music. J. 31(4), 47–58 (2007)View ArticleGoogle Scholar
- A Smaill, G Wiggins, M Harris, Hierarchical music representation for composition and analysis, computers and the humanities. Kluwer Acad. Publishers 27(1), 7–17 (1993)Google Scholar
- X Serra, The Musical Communication Chain and Its Modeling, Mathematics and Music, 243–255, 2002Google Scholar
- M Besson, AD Friederici, Language and music: a comparative view. Music. Percept. 16(1), 1–9 (1998)View ArticleGoogle Scholar
- JH Lee, JS Downie, A Renear, Representing Korean traditional music notation in XML, in Proceedings of the Third International Conference on Music Information Retrieval. (IRCAM Centre Pompidou, Paris, 2002)Google Scholar
- AD Patel, Music and the brain: three links to language. Oxford Handbook of Music Psychology. (2008). doi:10.1093/oxfordhb/9780199298457.013.0019Google Scholar
- F Ramus, M Nesport, J Mehler, Correlates of lingustic rhythm in the speech signal. Cognition 73, 265–292 (1999)View ArticleGoogle Scholar
- J Kippen, B Bel, Modelling music with grammars: formal language representation in the Bol Processor, Computer Representations and Models in Music. (Academic Press Limited, 1992), pp. 207–232Google Scholar
- F Lerdahl, R Jackendoff, A generative theory of tonal music (MIT Press, Cambridge, 1985), pp. 130–162Google Scholar
- R F Erickson, The DARMS project: A status report, Computing and the Humanities. Springer 9(6), 291–298 (1975)Google Scholar
- M Kuuskankare, M Laurson, Expressive notation package. Comput. Music. J. 30(4), 67–79 (2006)View ArticleGoogle Scholar
- MIDI Manufactures Association. Making music with MIDI, https://www.midi.org/. Accessed on 25 March 2015
- Michael Good, MusicXML. http://www.musicxml.com/. Accessed on 1 April 2015
- D Politis, D Margounakis, S Lazaropoulos, Leontios, Papaleontiou, George Botsaris, Konstantinos Vandikas, Emulation of ancient Greek music using sound synthesis and historical notation. Comput. Music. J. 32(4), 48–63 (2008)View ArticleGoogle Scholar
- P Chordia, A system for the analysis and representation of Bandishes and Gats using Humdrum syntax, in Proceedings of the 2007 Frontiers of Research in Speech and Music Conference (Mysore, India, 2007)Google Scholar
- MA Weiss, Data Structures and Algorithm Analysis in C++, Pearson Education, 1993Google Scholar
- Creative Commons. http://creativecommons.org/licenses/by-nc-sa/2.0/fr/legalcode. Accessed on 28 Dec 2015