Executive Summary : | The project aims to explore the computational aspect of Indian classical music for investigating the formation of Raag. The theory of Raag and incorporation of Alankars (ornamentation of note) in the Raag are the predominant factors in Indian classical music. A Raag has pre-specification regarding its features such as Pakad (catch phrases), Arohan and Avarohan (rule for ascending and descending), Vadi (most significant note) and Samvadi (second most significant note) etc. The study aims to analyze and validate the role of these traditional pre-specified features in forming a Raag and explore the possibility for additional features. Indian classical music also has Alankars (ornamentation of notes) such as Gamak, Murki, Meend, and Khatka that are rendered while artists perform any Raag. The study also aims to investigate how these Alankars are harnessed by using different swars and how these Alankars are incorporated by the artist to create a distinct form of a Raag. In this project, machine learning and soft computing techniques are to be explored in order to understand the formation of a Raag from the perspective of both pre-specified features of Raag and incorporation of Alankars in Raag. In particular, the neural network models such as deep learning, extreme machine learning and convolution neural networks are to be used for extracting features associated with Raag formation, while fuzzy logic and intelligent computing techniques are to be used to handle fuzziness of features.
The performances of various artists are to be collected from different online and offline sources to prepare databases for the experimental analysis. The audio or audio-visual recordings available in these sources may not be in the form required for the experiments. Moreover, the sound quality and diversity of available recordings are the two major issues. A lot of preprocessing is required to prepare currently available recordings in the form required for experiments, which is supposedly a time-consuming process. Therefore, recording of different artists as per the need of the experiments is to be done and datasets are to be prepared accordingly. This alternative also has a potential risk of unavailability of studio recording facilities and artists for recordings. On successful completion of the project, overcoming the potential risks associated with database preparation, a Raag analyzer software package and mobile apps related to music analysis are to be delivered. The software package will provide a platform for analyzing Raag and generating results for empirical validations. The software package will be made available online for the researchers of other institutes interested in Indian classical music. The mobile apps will help understanding the Raag and guiding the newbies to Raag. The apps will also helpful for vocalist as well as instrumentalist who are already familiar with Raag to verify and enhance their grip by in-depth understanding of Raag. |