FINITE STATE MODELS FOR GENERATION OF HINDUSTANI CLASSICAL MUSIC

We take a look at a model of computer aided generation of Hindustani Classical music. The system implemented using the model is essentially a simulation of the ‘Arohana’ and ‘Avarohana’ of a particular ‘Raga’. The artificial generator creates note sequences using hand-crafted bigram finite state models which conform to the grammar of a Raga. We also describe a procedure that intends to measure the quality of the artificially generated compositions by comparing them with random sequences which have been generated from unigram models. The measure makes it evident that the computer generated sequences are better than the random ones. However, the compositions are nowhere near the quality of human created ones and therefore this leaves a lot of scope of further research in similar areas.