RAGA IDENTIFICATION
        IN
  CARNATIC MUSIC




           G. Varsha Bargavi
              CSE, 3rd Year
We will be discussing…
           Introduction
    1      What automatic Raga identification can provide


    2     Characteristics of Carnatic music
          And Related Terminologies


    3     Existing work
          How automation is not complete


    4     System Architecture – The various modules
          Detailed explanation of every phase

    5     Conclusion
          Acknowledgements and references
1.Introduction
What automatic Raga identification can provide

• Computational Musicology                       Music Databases, Music
                                                 Analysis, Artificial
                                                 Production of Music

• Music Information Retrieval
                                                 Classification, clustering,
                                                 identification,
• Raga Identification                            recognition, perception,
                                                 cognition, affect,
                                                 emotions, evaluation

         as a basis for music search
         for generating playlists based on themes

        for evaluating performance and accuracy of raga

         for answering queries regarding ragas
2. Characteristics of Carnatic Music
  And related terminologies…

  • Swara (note)
  • Octave
  • Raga
  • Arohanam and
    Avarohanam
    (swara sequence)
  • Graha Bedham /Shruti
    Bedham
Graha Bedham/Shruti Bedham

Graha = position

Bedham = shift

 Process (or result of the process) of
shifting the Tonic note (śruti) to another
note in the raga and arriving at a different
raga.
Śrut
               Mela i
   Rāgam        # Toni
                        C    D    E   F    G    A    B    C    D     E     F     G    A     B     C
                    c

Shankarabhara
              29     C   S   R2   G3 M1    P    D2   N3   S'   R2'   G3'   M1'   P'   D2'   N3'   S' '
    nam

Karaharapriya 22     D       S    R2 G2    M1   P    D2   N2   S'


 Hanumatodi    08    E            S   R1   G2   M1   P    D1   N2    S'

   Kalyani     65    F                S    R2   G3   M2   P    D2    N2    S'

Harikambhoji   28    G                     S    R2   G3   M1   P     D2    N2    S'


Natabhairavi    21   A                          S    R2   G2   M1    P     D1    N2   S'


   Invalid
                --   B                               S    R1   G2    M1    M2    D1   N2    S'
  Melakarta

Shankarabhara
              29     C   S   R2   G3 M1    P    D2   N3   S'   R2'   G3'   M1'   P'   D2'   N3'   S' '
    nam
3. Existing Work
 • Swara Identification and singer identification in Carnatic music


 • Tansen- Hindustani music raga identification


 •   Melody retrieval and song detection in Western music

 Limitations…
 • Some of the work do not consider polyphonic music signals


 • Not fully automated
     Fundamental frequency(tonic note) - given as input manually
4.System Architecture
Input
audio     Signal            Segmentation          Frequency
        Separation                                Extraction



         Output            Raga Mapping        Swara sequences
                                                construction




           Raga database              Finite Automata
Signal Separation
Segmentation and Frequency
Extraction


 F
 r
 e
 q
 u
 e
 n
 c
 y




             Time
Swara Sequences Construction
Applying graha bedham..



                          Invalid
                          sequence
Finite Automata

       To eliminate invalid sequences



 S .... M1 M2 .... S’          Invalid!

  M1            M2         ε
           q1         q2         q3
Bayesian Network Model
Simple example-                      In Raga Detection
                                             a1           a2           a3
                                        X1          X2                      X3

                                              b21              b32
                                                         b22
                                      b11                                        b33
                                        Y1           Y2                Y3


                         Xi – Phrases/Gamakas (Random Variables)
                         Yi- Raga
                         ai- Probability associated with Xi
                         bij- conditional probability of yj given Xi

    Identified raga = Raga ‘Yi’ with highest probability Value
5. Conclusion
Carnatic Music Ragas- Variables + Constants

Constants (Swaras) - distinctly specified using
Automata

Variables (Phrases) - Represented using associated
probabilities and mapped to respective Raga using a
probabilistic (Bayesian) network model
iRaga
   Imagine...

   The next version of iPhone having an app
   called iRaga

   It can list songs of your favourite raga ...

   And even create playlists based on a type
   of emotion , or your mood!
References
1. Prof. Sambamurthy, South Indian Music, vol. 4, The Indian
Music Publishing House, Madras, India, 1982

2. Rajeswari Sridhar and T.V. Geetha, “Swara identification of Carnatic music”, IEEE
Computer Society press, proceeding of ICIT 2006

3. Rajeswari Sridhar and T.V. Geetha, “Music Information
Retrieval of Carnatic Songs Based on Carnatic Music Singer
Identification”, IEEE conference on Computer Science, ICCEE 2008

4. Gaurav Pandey, et al “Tansen: A system for automatic Raga identification” IICAI,
2003

Links-
http://www.rasikas.org/
http://news.cnet.com/2009-1001-984695.html
http://www.it.iitb.ac.in
http://en.wikipedia.org/wiki/Graha_bedham
Acknowledgements
• Mrs. N. Mythili Jagannathan


• Mr. S. Rangarajan


• Mrs. Uma Maheshwari
Raga Identification In Carnatic Music
Raga Identification In Carnatic Music

Raga Identification In Carnatic Music

  • 1.
    RAGA IDENTIFICATION IN CARNATIC MUSIC G. Varsha Bargavi CSE, 3rd Year
  • 2.
    We will bediscussing… Introduction 1 What automatic Raga identification can provide 2 Characteristics of Carnatic music And Related Terminologies 3 Existing work How automation is not complete 4 System Architecture – The various modules Detailed explanation of every phase 5 Conclusion Acknowledgements and references
  • 3.
    1.Introduction What automatic Ragaidentification can provide • Computational Musicology Music Databases, Music Analysis, Artificial Production of Music • Music Information Retrieval Classification, clustering, identification, • Raga Identification recognition, perception, cognition, affect, emotions, evaluation  as a basis for music search  for generating playlists based on themes for evaluating performance and accuracy of raga  for answering queries regarding ragas
  • 4.
    2. Characteristics ofCarnatic Music And related terminologies… • Swara (note) • Octave • Raga • Arohanam and Avarohanam (swara sequence) • Graha Bedham /Shruti Bedham
  • 6.
    Graha Bedham/Shruti Bedham Graha= position Bedham = shift Process (or result of the process) of shifting the Tonic note (śruti) to another note in the raga and arriving at a different raga.
  • 7.
    Śrut Mela i Rāgam # Toni C D E F G A B C D E F G A B C c Shankarabhara 29 C S R2 G3 M1 P D2 N3 S' R2' G3' M1' P' D2' N3' S' ' nam Karaharapriya 22 D S R2 G2 M1 P D2 N2 S' Hanumatodi 08 E S R1 G2 M1 P D1 N2 S' Kalyani 65 F S R2 G3 M2 P D2 N2 S' Harikambhoji 28 G S R2 G3 M1 P D2 N2 S' Natabhairavi 21 A S R2 G2 M1 P D1 N2 S' Invalid -- B S R1 G2 M1 M2 D1 N2 S' Melakarta Shankarabhara 29 C S R2 G3 M1 P D2 N3 S' R2' G3' M1' P' D2' N3' S' ' nam
  • 8.
    3. Existing Work • Swara Identification and singer identification in Carnatic music • Tansen- Hindustani music raga identification • Melody retrieval and song detection in Western music Limitations… • Some of the work do not consider polyphonic music signals • Not fully automated Fundamental frequency(tonic note) - given as input manually
  • 9.
    4.System Architecture Input audio Signal Segmentation Frequency Separation Extraction Output Raga Mapping Swara sequences construction Raga database Finite Automata
  • 10.
  • 11.
  • 12.
  • 13.
    Applying graha bedham.. Invalid sequence
  • 14.
    Finite Automata To eliminate invalid sequences S .... M1 M2 .... S’ Invalid! M1 M2 ε q1 q2 q3
  • 15.
    Bayesian Network Model Simpleexample- In Raga Detection a1 a2 a3 X1 X2 X3 b21 b32 b22 b11 b33 Y1 Y2 Y3 Xi – Phrases/Gamakas (Random Variables) Yi- Raga ai- Probability associated with Xi bij- conditional probability of yj given Xi Identified raga = Raga ‘Yi’ with highest probability Value
  • 16.
    5. Conclusion Carnatic MusicRagas- Variables + Constants Constants (Swaras) - distinctly specified using Automata Variables (Phrases) - Represented using associated probabilities and mapped to respective Raga using a probabilistic (Bayesian) network model
  • 17.
    iRaga Imagine... The next version of iPhone having an app called iRaga It can list songs of your favourite raga ... And even create playlists based on a type of emotion , or your mood!
  • 18.
    References 1. Prof. Sambamurthy,South Indian Music, vol. 4, The Indian Music Publishing House, Madras, India, 1982 2. Rajeswari Sridhar and T.V. Geetha, “Swara identification of Carnatic music”, IEEE Computer Society press, proceeding of ICIT 2006 3. Rajeswari Sridhar and T.V. Geetha, “Music Information Retrieval of Carnatic Songs Based on Carnatic Music Singer Identification”, IEEE conference on Computer Science, ICCEE 2008 4. Gaurav Pandey, et al “Tansen: A system for automatic Raga identification” IICAI, 2003 Links- http://www.rasikas.org/ http://news.cnet.com/2009-1001-984695.html http://www.it.iitb.ac.in http://en.wikipedia.org/wiki/Graha_bedham
  • 19.
    Acknowledgements • Mrs. N.Mythili Jagannathan • Mr. S. Rangarajan • Mrs. Uma Maheshwari