A Study on the Structure and Practices of Vakra Scales in Carnatic Music

Abstract

One of the world's oldest and most hierarchically structured classical music systems, Carnatic music is based on a hierarchical tonal system, and the concept of raga is the central vehicle of melodic expression. In this system, the scales with a non-linear, zigzag note sequence, known as vakra, play a crucial role: They give ragas a distinctive melodic contour that poses unique challenges for musical analysis and computational rag recognition. The present paper reviews the sixteen peer-reviewed articles and conference proceedings published till 2023, to explore the structural characteristics, categorical classification, performance practices and computational modelling issues of vakra scales in the context of Carnatic music. Key quantitative results confirm that the recognition accuracy of vakra ragas ranges from 68% (using pitch-distribution technique (Koduri et al., 2012)) to 89% (using joint deep-learning models (Bhat et al., 2023)), while the recognition accuracy of krama ragas falls in the range from 79% to 94%, indicating a persistent accuracy gap between 5%–11% due to the sequence-dependent nature of vakra patterns. The main empirical source used in recent studies is the Saraga open data set (Srinivasamurthy et al., 2021), which includes 249 Carnatic recordings over 72 ragas, of which around 38% have vakra properties. It is also shown that music therapy is clinically relevant in therapeutic research (Krishna et al., 2022) where anxiety was found to reduce statistically significantly (p < .05) in music therapy applications. For the representation of vakra sequences in machine-readable format, computational notation systems like iSargam (Mammen et al., 2016) have played a key role. The paper finishes with a discussion of open challenges and future research directions in vakra scale research, such as the creation of gamaka-aware modelling, the extension of annotated datasets and comparative analysis across traditions.