Automatic Synthesis of Notes Based on Carnatic Music Raga Characteristics

  • Janani VaradharajanEmail author
  • Guruprasad Sridharan
  • Vignesh Natarajan
  • Rajeswari Sridhar
Part of the Smart Innovation, Systems and Technologies book series (SIST, volume 27)


In this paper, we propose two methods to automatically generate notes (Swaras) conforming to the rules of Carnatic Music, for a Raga of the user’s choice as the input. The proposed methods are purely statistical in nature. The system requires training examples for learning the probability model of the chosen Raga and no hand-coded rules are required. Hence, it is easy to extend this method to work with a large number of Ragas. Each proposed method involves a Learning Phase and Synthesis Phase. In the Learning Phase, an already existing composition of the desired Raga is used to learn the transition probabilities between swara sequences based on the Raga lakshana characteristics. In the Synthesis Phase, using the previously constructed transition table, swaras are generated for the desired Raga. We describe two methods - one based on First Order Markov Models and the other based on Hidden Markov Models. We also provide comparison of the performance of both the approaches based on feedback from Carnatic experts.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    Sambamoorthy, P.: South Indian Music, vol. 4. Indian Music Publishing (1969)Google Scholar
  2. 2.
    Sahasrabuddhe, H.V.: Analysis and Synthesis of Hindustani Classical Music, University of Poona (1992),
  3. 3.
    Das, D., Choudhury, M.: Finite state models for generation of Hindustani classical music. In: Proceedings of International Symposium on Frontiers of Research in Speech and Music (2005)Google Scholar
  4. 4.
    Subramanian, M.: Generating Computer Music from Skeletal Notations for Carnatic Music Compositions. In: Proceedings of the 2nd CompMusic Workshop (2012)Google Scholar
  5. 5.
    Hill, S.: Markov Melody GeneratorGoogle Scholar
  6. 6.
    Steinsaltz, D., Wessel, D.: In Progress,The Markov Melody Engine: Generating Random Melodies With Two-Step Markov Chains. Technical Report, Department of Statistics, University of California at BerkeleyGoogle Scholar
  7. 7.
    Kohlschein, C.: An introduction to hidden Markov models: Probability and Randomization in Computer Science. Aachen University (2006-2007)Google Scholar
  8. 8.
    Pedregosa, F., Varoquaux, G., Gramfort, A., Michel, V., Thirion, B., Grisel, O., Blondel, M., Prettenhofer, P., Weiss, R., Dubourg, V., Vanderplas, J., Passos, A., Cournapeau, D., Brucher, M., Perrot, M., Duchesnay, E.: Scikit-learn: Machine learning in Python. The Journal of Machine Learning Research 12, 2825–2830 (2011)zbMATHGoogle Scholar
  9. 9.
    Oliphant, T.E.: Python for scientific computing. Computing in Science & Engineering 9(3), 10–20 (2007)CrossRefGoogle Scholar

Copyright information

© Springer International Publishing Switzerland 2014

Authors and Affiliations

  • Janani Varadharajan
    • 1
    Email author
  • Guruprasad Sridharan
    • 1
  • Vignesh Natarajan
    • 1
  • Rajeswari Sridhar
    • 1
  1. 1.Department of Computer Science and Engineering, College of EngineeringAnna UniversityChennaiIndia

Personalised recommendations