Rāgawise: A Lightweight Real-time Rāga Recognition System for Indian Art Music


We demonstrate a web-based lightweight real-time melodic analysis and visualization system for Indian art music. Our system uses pitch class profiles, pitch transitions and melodic phrases for melodic characterization and rāga recognition. For each rāga we store a dictionary of its svaras (notes), svara transitions, and typical melodic phrases. We process the input vocals in real-time to estimate pitch, and subsequently perform melody transcription. The likelihood of each rāga is updated in real-time based on the transcribed melody. In order to highlight the melodic events that are characteristic of a rāga, we perform a dynamic visualization of the evolution of the likelihood of all the rāgas for the sung melodic excerpt.

A short paper describing the system can be found here.


Ragawise (Beta): TRY IT YOURSELF (Use Chrome or Firefox)

Screenshot of the beta version of the web-interface for Ragawise.


The development of this system started as a hack that was presented in HAMR'2015 in ISMIR in Malaga. We also won 'Best-Code' award for this hack. There is also a wikipage describing the hack.

We use the YIN algorithm to estimate the fundamental frequency in real-time. We have implemented YIN in javascript (YIN.js), the code for which is openly available here.


Here is a short video that demonstrates ragawise: