Simulating Raga Notes with a Markov Chain of Order 1-2

TL;DR

This paper introduces an extended semi-natural algorithmic composition method using first and second order Markov chains to simulate Raga Bageshree notes, aiding composers in music sequence generation.

Contribution

The paper proposes the SNCA2 algorithm, extending existing methods with second order Markov chains for more nuanced music note sequence simulation.

Findings

Second order Markov chain better captures note transitions.

Higher order Markov chains (>2) are less effective due to sparse matrices.

The approach assists composers in choosing between first and second order models.

Abstract

Semi Natural Algorithmic composition (SNCA) is the technique of using algorithms to create music note sequences in computer with the understanding that how to render them would be decided by the composer. In our approach we are proposing an SNCA2 algorithm (extension of SNCA algorithm) with an illustrative example in Raga Bageshree. For this, Transition probability matrix (tpm) was created for the note sequences of Raga Bageshree, then first order Markov chain (using SNCA) and second order Markov chain (using SNCA2) simulations were performed for generating arbitrary sequences of notes of Raga Bageshree. The choice between first and second order Markov model, is best left to the composer who has to decide how to render these music notes sequences. We have confirmed that Markov chain of order of three and above are not promising, as the tpm of these become sparse matrices.