Machine Learning with Evolutionary Parameter Tuning for Singing Registers Classification - Signals 2025 - 9

Aug 5, 2026

Upload by Quinn

Human voice production is driven by a complex biological mechanism that generates and modulates sound. Prior studies have applied machine-learning methods to analyze singing-voice characteristics, yet reported classification performance indicates room for improvement. Further gains can be achieved by introducing under-utilized optimization strategies. This work proposes differential evolution optimization of hyperparameters across three selected ML models to classify chest, mixed, and head registers, using 350 audio files and TSFEL feature extraction, achieving 97.60% average accuracy.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages22

Reading Time55 min

AccessFree to download

Introduction

Voice production and the source-filter model

Proposed approach

Differential Evolution hyperparameter tuning

Selected ML models for register classification

Data and feature extraction

Dataset construction (350 audio files)

TSFEL temporal feature extraction

Results and discussion

Extreme Gradient Boosting optimized with DE

Conclusion

Article Machine Learning with Evolutionary Parameter Tuning for Singing Registers Classification Tales Boratto 1, Gabriel de Oliveira Costa 2, Alexsandro Meireles 3, Anna Klara Sá Teles Rocha Alves 4, Camila M. Saporetti 5, Matteo Bodini 6, *, Alexandre Cury 7 and Leonardo Goliatt 7 Academic Editors: Alexander Kocianand Constantine Kotropoulos Received: 31 December 2024 Revised: 4 February 2025 Accepted: 18 February 2025 Published: 21 February 2025 Citation: Boratto, T.; Costa, G.d.O.; Meireles, A.; Alves, A.K.S.T.R.; Saporetti, C.M.; Bodini, M.; Cury, A.; Goliatt, L. Machine Learning with Evolutionary Parameter Tuning for Singing Registers Classification. Signals 2025, 6, 9. [https://doi.org/](https://doi.org/)[ ](https://doi.org/)[10.3390/signals6010009](10.3390/signals6010009) Copyright: © 2025 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license ([https://creativecommons.org/](https://creativecommons.org/)[ ](https://creativecommons.org/)[licenses/by/4.0/](licenses/by/4.0/)) . 1 Graduate Program in Computational Modeling, Federal University of Juiz de Fora, Juiz de Fora 36036-900, MG, Brazil; [tales.boratto@estudante.ufjf.br](tales.boratto@estudante.ufjf.br) 2 Department of Mechatronics Engineering, Federal Institute of Southeast Minas Gerais, Juiz de Fora 36080-001, MG, Brazil; [gabrieloliveira@grupocsc.com.br](gabrieloliveira@grupocsc.com.br) 3 Department of Languages and Literature, Federal University of Espírito Santo, Vitória 29075-910, ES, Brazil; [meirelesalex@gmail.com](meirelesalex@gmail.com) 4 Graduate Program in Nursing, Federal University of Juiz de Fora, Juiz de Fora 36036-900, MG, Brazil; [alves.anna@estudante.ufjf.br](alves.anna@estudante.ufjf.br) 5 Department of Computational Modeling, Polytechnic Institute, Rio de Janeiro State University, Nova Friburgo 22000-900, RJ, Brazil; [camila.saporetti@iprj.uerj.br](camila.saporetti@iprj.uerj.br) 6 Dipartimento di Economia, Management e Metodi Quantitativi, Università degli Studi di Milano, Via Conservatorio 7, 20122 Milano, Italy 7 Department of Computational and Applied Mechanics, Federal University of Juiz de Fora, Juiz de Fora 36036-900, MG, Brazil; [alexandre.cury@ufjf.br](alexandre.cury@ufjf.br) (A.C.); [leonardo.goliatt@ufjf.br](leonardo.goliatt@ufjf.br) (L.G.) * [Correspondence: matteo.bodini@unimi.it](Correspondence: matteo.bodini@unimi.it) Abstract: Behind human voice production, a complex biological mechanism generatesand modulates sound. Recent research has explored machine-learning (ML) techniques to analyze singing-voice characteristics. However, the classification efficiency reported in such research works suggests the possibility of improvement. In addition, there is also scope for further improvement through the application of still under-utilized optimization techniques. Thus, the present article proposes a novel approach that leverages the Differential Evolution (DE) algorithm to optimize hyperparameters within three selected ML models, with the aim of classifying singing-voice registers i.e., chest, mixed, and head registers) . To develop the present study, a dataset of 350 audio files encompassing the three aforementioned registers was constructed. Then, the TSFEL Python library was employed to extract 14 pieces of temporal information from the audio signals for subsequent classification by the employed ML models. The obtained findings demonstrated that the Extreme Gradient Boosting model, optimized with DE, achieved an average classification accuracy of 97.60%, thus indicating the efficacy of the proposed approach for singing-voice register classification. Keywords: singing registers; machine learning; differential evolution; classification; optimization 1. Introduction Behind the apparent simplicity of the sound of the human voice lies an intricate interplay of resonances

FAQs

What is the main goal of the proposed study?

To improve singing-voice register classification by optimizing ML model hyperparameters using the Differential Evolution algorithm.

Which singing registers are classified in the work?

The approach classifies chest, mixed, and head singing registers.

How were audio features extracted for the machine-learning models?

The study constructed a dataset of 350 audio files and used the TSFEL Python library to extract 14 temporal information features from the signals.

Aug 5, 2026

Upload by Quinn