Categorization of Tablas by Wavelet Analysis

TL;DR

This paper uses wavelet analysis to examine spectral features of tabla strokes, enabling the categorization of different tablas based on their harmonic and frequency patterns.

Contribution

It extends previous spectral modeling work by applying wavelet transform to analyze and classify tabla strokes from multiple instruments.

Findings

Wavelet analysis reveals distinct spectral patterns for different tabla strokes.

Distribution of dominant frequencies helps differentiate tablas.

Harmonic behavior varies across sub-bands and tablas.

Abstract

Tabla, a percussion instrument, mainly used to accompany vocalists, instrumentalists and dancers in every style of music from classical to light in India, mainly used for keeping rhythm. This percussion instrument consists of two drums played by two hands, structurally different and produces different harmonic sounds. Earlier work has done labeling tabla strokes from real time performances by testing neural networks and tree based classification methods. The current work extends previous work by C. V. Raman and S. Kumar in 1920 on spectrum modeling of tabla strokes. In this paper we have studied spectral characteristics (by wavelet analysis by sub band coding method and using torrence wavelet tool) of nine strokes from each of five tablas using Wavelet transform. Wavelet analysis is now a common tool for analyzing localized variations of power within a time series and to find the frequency distribution in time frequency space. Statistically, we will look into the patterns depicted by harmonics of different sub bands and the tablas. Distribution of dominant frequencies at different sub-band of stroke signals, distribution of power and behavior of harmonics are the important features, leads to categorization of tabla.