Brain- Computer Interface for Carnatic Rāga Identification: New Directions and Prospects Aparna S. Narayan Abstract Background: The study of Rāga identification in Carnatic music is a fascinating, interdisciplinary field of study with strong associations to brain functioning, mood regulation, and cognitive psycho-auditory perception. Rationale: Although research into this area has been carried out, contemporary scientists have focused more on Hindustani music rather than its Carnatic counterpart. This study was undertaken primarily to consolidate current research trends in this area from diverse perspectives- i.e, to combine research by physicists, neuroscientists, computer scientists, electrical engineers, musicians, etc. Strategy: To carry out this review a preliminary study of the basic concepts in Carnatic music such as shruti, talā, nāda, swara, melakartā rāga, janya rāga, vakra rāga etc. Following this, an exhaustive literature survey was done using the keywords/phrases ‘rāga and brain’, ‘rāga identification’, ‘Carnatic music’, ‘Hindustani music’ etc. Over 50 papers were consulted for this review, most of which were published in reputed journals such as - Physica A, Cognitive neurodynamics, Music cognition, and Audio and speech processing Results: After the literature survey, the merits and demerits of the papers were summarised and a new multi-disciplinary approach to identify rāgas has been outlined based on these findings. This new approach attempts to combine neuroscientific aspects such as EEG analysis with electronic/signal processing research into constructing neural networks based on acoustic features. Further, future directions and advantages from this research have been highlighted. Conclusions: Through this paper, new directions for interdisciplinary research into carnatic rāga identification are summarised. However, the experimental validation of the proposed theoretical model is yet to be undertaken.
Rationale Robust and fascinating research notwithstanding, there are at least three major deficiencies in the efforts to scientifically identify rāga. The first of which was a . disproportionately less amount of research into Carnatic music in comparison to the Hindustani style. Many papers implied that research methods employed to categorize Hindustani rāgas could also be used to do the same in Carnatic music. Although the broad techniques are employable for both genres, there are fundamental differences in the derivation and rendering of the rāga between the two styles. The second inadequacy is related to the idea of ‘gamakas’. Characterizing gamakas has been a particular challenge and scientific work related to this, in particular, is sparse. Lastly, there has been a distinct lack of coordination between the neuroscientific and the physical/computer science branches of research. Neuroscience-directed research is very important to accurately identify rāgas as many rāgas have the same notes and are differentiated based on the emotional responses they evoke. Further, current neuroscientific work can be enriched by in-depth work into the rāga associated wave functions by physicists and computer scientists
Introduction Carnatic music, one of the two major subgenres of Indian Classical music, commonly associated with the southern states of India has been theorized to have split from its northern counterpart sometime in the 14th century. The rāga, which is an array of systematic melodic structures akin to the western ‘mode’ is the central component around which this music genre is based. The term rāga literally means ‘coloring’ and is considered to be able to significantly alter moods or ‘color the mind’. There are over 300 individual rāgas, each of which is accompanied by its own characteristic emotional response or rasa. Several musicians, since time immemorial have actively engaged in research to characterize and identify the signature of each rāga. Modern-day physicists, computer scientists, and electrical engineers have attempted to expand into this area by attempting to understand each rāga’s frequency, wave function, pitch modulations etc. Parallelly, neuroscientists have investigated the response of different brain areas to different rāgas and mathematically modelled these observations.
Methodology Over 50 papers were consulted to understand the various aspects of research into rāga identification. Their methodology was studied in-depth to analyze current directions of research, common challenges in this area etc. Based on this, an integrated methodology has been summarized in a flowchart based infographic.
Results
Figure 1: Chart on the classification of the 72 Melakarta Ragas. Image taken from Wikimedia Commons. No copyright infringement intended. Basavarajtalwar / CC BY-SA (https://creativecommons.org/licenses/by-sa/3.0)
References Rao, P. Swapna, and A. Soumya. "A Study on Music based Audio and Brain Signal Processing." 2019 4th International Conference on Computational Systems and Information Technology for Sustainable Solution (CSITSS). Vol. 4. IEEE, 2019.; Banerjee, Archi, et al. "EVIDENCE OF HYSTERESIS FROM THE STUDY OF NON LINEAR DYNAMICS OF BRAIN WITH MUSIC STIMULI."; Asokan, Akshaya, K. Gunavathi, and R. Anitha. "Classification of Melakartha ragas using neural networks." 2017 International Conference on Innovations in Information, Embedded and Communication Systems (ICIIECS). IEEE, 2017; .Ranjani H, G., and Thippur V. Sreenivas. "Raga identification using repetitive note patterns from prescriptive notations of Carnatic music." arXiv (2017): arXiv-1711.
Interesting conclusions from past research: Research undertaken by Sanyal et al showed that there are possible long-term effects of rāgas on the brain. When the acoustic stimulus was removed, the long alpha waves observed for the rāga played were retained for 77-120 seconds. Further they also noted that ‘sad’ emotional responses associated with the right frontal lobe are more pronounced than ‘happy’ emotions recorded in the left frontal lobe. Research Challenges identified: (1) Characterizing ‘gamakas’ or movement around the notes (pitch inflection) often leads to the identification of wrong notes (2) Results may vary drastically with the response to the different timbers of different musicians and instruments (3) Many papers concentrated on contrasting rasas or extreme emotions without considering the spectrum of rāgas that fall in between these extremes (4) Vakra Rāgas which don’t have uniform arohana and avarohana and have repeating notes are difficult to identify. Scope: (1) Development of software to identify rāgas rendered by vocalists and musicians; Understanding musical preferences and recommending personalized lists of new music (2) A renewed understanding of how our brain identifies and processes audio signals (3) Effect of music in mood regulation and long-term effects of music on the brain (4) New directions into psychoacoustics and research into how our brain recognizes patterns in music (5) Due to the highly systematic nature of carnatic music robust research into this area may provide a solid foundation and framework to classify the psychoacoustic of other musical genres (6) Music and pattern recognition may also be a valuable tool to evaluate the brain’s functioning (7) May provide information on the differences between acoustic processing between trained-musicians and untrained musicians.