Design of Formal Grammar and Model for Composition of Indian Music

Abstract

Music is considered a universal language that persists everywhere. Indian music is very renowned due to its sophistication and rhythmic diversities. Music composition and mathematical computation are closely related and have similar aesthetics. In the last few years, musicians and researchers have used a computer to design compositional models for musical patterns. Computer-assisted music composition has been an active research area since mid-1900s, and the potential of computer systems in contributing music-related research had been imagined as a future possibility in 1950. Computational musicology is an interdisciplinary area that includes the contribution of music and computer science methods. This thesis explores the intricate world of Indian music composition through formal grammar and modeling techniques. The primary objective is to create a framework that captures the essence of Indian music composition while providing a platform for innovation and automation in music creation. This thesis proposed a technique for generating musical sequence to help musicians as well as non-musicians compose musical structures. The work aims to generate formal grammar for an Indian musical composition. A musical tool is used to generate a musical sheet for the composition used. We outline musical rules and formal grammar rules used for music composition. The probabilistic context-free grammar is modeled to generate the same progression as the composition’s. Variable Order and Gapped hidden Markov model for unstructured elements can capture variable length dependencies with variable gaps in sequential data. VOGUE uses a Variable Gap Sequence miner algorithm to extract frequent patterns in a sequence with variable gaps. In this thesis, we applied the VOGUE model to design the musical sequence of notes in bandish of raga Bhairav, a classical Indian music. Furthermore, we analyzed the benefits of VOGUE model over the standard HMM. Tala of Indian music represents rhythmic aspects in musical compositions. This thesis used formal grammar to examine the patterns in Tala’s kaidas and paltas. Results indicate that the patterns in the paltas of Tala’s kaida exhibit cross-serial dependencies, which context-free grammar cannot represent. Further, we represented these patterns using the deep pushdown automata mathematical model. The proposed technique can be applied to check the correctness of the patterns used in Tala’s Kaidas and Paltas.