Indian Music Generation and Analysis using LSTM

Main Article Content

Jyoti Lele, Aditya Abhyankar

Abstract

While writing a song with a distinctive melody and appropriate chords for accompaniment, a musician must experiment a lot.  This work describes a machine-learning technique that uses LSTM to automatically generate chords and note sequences from a collection of Indian songs that are stored as .wav files. We propose to employ machine learning as a tool to manipulate and understand massive amounts of data generated from .wav files of various Indian songs. A brief introduction about music and its components is provided in the paper which includes notes, instrument, chords and durations. This analysis is very useful while generating sequestered musical constructs to synthesize Indian songs of any language.

Article Details

Section
Articles