A Computational Analysis of Pitch Drift in Unaccompanied Solo Singing using DBSCAN Clustering

TL;DR

This paper introduces a computational method to measure pitch drift in unaccompanied singing performances using pitch histograms and DBSCAN clustering, providing insights into how performers' skill and performance length influence pitch stability.

Contribution

The paper presents a novel approach combining pitch histograms and DBSCAN clustering to quantify pitch drift in solo singing performances.

Findings

Effective measurement of pitch drift using the proposed method

Correlation between performer skill and pitch stability

Analysis of pitch drift over performance duration

Abstract

Unaccompanied vocalists usually change the tuning unintentionally and end up with a higher or lower pitch than the starting point during a long performance. This phenomenon is called pitch drift, which is dependent on various elements, such as the skill of the performer, and the length and difficulty of the performance. In this paper, we propose a computational method for measuring pitch drift in the course of an unaccompanied vocal performance, using pitch histogram and DBSCAN clustering.