Automatic detection of Parkinson’s disease from speech signals using the Fourier–Bessel domain adaptive wavelet transform
Nayak SS, Darji AD, Shah PK, Orozco-Arroyave JR (2026)
Publication Type: Journal article
Publication year: 2026
Journal
Book Volume: 245
Article Number: 111217
DOI: 10.1016/j.apacoust.2025.111217
Abstract
There has been a growing interest in the development of automated methods to diagnose Parkinson’s disease from speech. These approaches can potentially be used in telemonitoring health applications; however, there is still much to be done in the process of developing accurate methods to perform the diagnosis. The purpose of this paper is to present a novel and efficient approach for detecting Parkinson’s disease from speech signals. Parkinson’s disease speech is modeled utilizing the Fourier–Bessel domain adaptive wavelet transform. The signal is decomposed by the Fourier–Bessel domain adaptive wavelet transform into several modes. Energy, entropy, increment entropy, and spectral entropy are extracted from each of the decomposed signals, and a combination of these features is evaluated using the isolated words and sustained vowels from the PC-GITA database. Support vector machine classifier achieves a maximum classification accuracy of 95 % for /drama/. Furthermore, with the aim of evaluating the generalization capability of the introduced approach, the model optimized with PC-GITA is used to perform the automatic classification of Parkinson’s disease vs. healthy control subjects in an independent dataset with a classification accuracy of 84 %. The results show that the proposed approach based on the Fourier–Bessel domain adaptive wavelet transform decomposition is accurate and efficient. Additionally, it showed robustness against unseen data collected under non-controlled acoustic conditions, making it a good candidate to develop computational systems that work properly in real-world clinical practice.
Involved external institutions
Sardar Vallabhbhai National Institute of Technology
India (IN)
Universidad de Antioquía (UDEA)
Colombia (CO)
India (IN)
Universidad de Antioquía (UDEA)
Colombia (CO)
How to cite
APA:
Nayak, S.S., Darji, A.D., Shah, P.K., & Orozco-Arroyave, J.R. (2026). Automatic detection of Parkinson’s disease from speech signals using the Fourier–Bessel domain adaptive wavelet transform. Applied Acoustics, 245. https://doi.org/10.1016/j.apacoust.2025.111217
MLA:
Nayak, Sudhansu Sekhar, et al. "Automatic detection of Parkinson’s disease from speech signals using the Fourier–Bessel domain adaptive wavelet transform." Applied Acoustics 245 (2026).
BibTeX: Download