Repertoire-Specific Vocal Pitch Data Generation for Improved Melodic Analysis of Carnatic Music

Abstract

Deep Learning methods achieve state-of-the-art in many tasks, including vocal pitch extraction. However, these methods rely on the availability of pitch track annotations without errors, which are scarce and expensive to obtain for Carnatic Music. Here we identify the tradition-related challenges and propose tailored solutions to generate a novel, large, and open dataset, the Saraga-Carnatic-Melody-Synth (SCMS), comprising audio mixtures and time-aligned vocal pitch annotations. Through a cross-cultural evaluation leveraging this novel dataset, we show improvements in the performance of Deep Learning vocal pitch extraction methods on Indian Art Music recordings. Additional experiments show that the trained models outperform the currently used heuristic-based pitch extraction solutions for the computational melodic analysis of Carnatic Music and that this improvement leads to better results in the musicologically relevant task of repeated melodic pattern discovery when evaluated using expert annotations. The code and annotations are made available for reproducibility. The novel dataset and trained models are also integrated into the Python package compIAM 1 which allows them to be used out-of-the-box.

Landscapes of coarticulation: The co-structuring of gesture-vocal dynamics in Karnatak music performance

“…For each of the 595 motifs, we extract a time series corresponding to the following features: the f0 of the predominant sung melody, using a machine learning methodology tailored for Karnatak music [52], also implemented as part of the compIAM package; Δf0, an approximation of the first derivative of the f0 curve as outlined in [53]; loudness,…”