@article{ART003300143},
author={Jeong Hun Park and Young-Kyoon Suh and Jeeyoung Kim},
title={Early Classification of Parkinson’s Disease Using Mel-Spectrogram Voice Analysis via Transfer Learning},
journal={Journal of The Korea Society of Computer and Information},
issn={1598-849X},
year={2026},
volume={31},
number={1},
pages={99-107}
TY - JOUR
AU - Jeong Hun Park
AU - Young-Kyoon Suh
AU - Jeeyoung Kim
TI - Early Classification of Parkinson’s Disease Using Mel-Spectrogram Voice Analysis via Transfer Learning
JO - Journal of The Korea Society of Computer and Information
PY - 2026
VL - 31
IS - 1
PB - The Korean Society Of Computer And Information
SP - 99
EP - 107
SN - 1598-849X
AB - Parkinson’s disease is a progressive neurodegenerative disorder, and early diagnosis is critical for slowing its progression. Voice changes are particularly notable among early indicators, offering a non-invasive pathway for timely detection. However, most existing approaches rely on traditional machine learning methods such as Support Vector Machines (SVM) and Support Vector Regression (SVR), which often fail to capture complex vocal patterns and thus exhibit limited generalization performance. This study proposes a voice-based diagnostic framework for Parkinson’s disease and related disorders. Voice recordings were transformed into Mel-spectrogram images and classified using deep learning models, including ResNet152V2 and DenseNet201. The dataset included not only Parkinson’s disease but also clinically similar conditions such as essential tremor, multiple system atrophy, and tau-Parkinsonism, alongside healthy controls. Experimental results show that our deep learning models achieve high accuracy (92%) in distinguishing Parkinson’s disease and related disorders from healthy individuals. These findings highlight the potential of voice-based deep learning approaches as a non-invasive, cost-effective tool for early diagnosis and clinical support in neurodegenerative disease management.
KW - Parkinson’s disease;Voice Signals;Mel-Spectrogram;Deep learning;Transfer learning
DO -
UR -
ER -
Jeong Hun Park, Young-Kyoon Suh and Jeeyoung Kim. (2026). Early Classification of Parkinson’s Disease Using Mel-Spectrogram Voice Analysis via Transfer Learning. Journal of The Korea Society of Computer and Information, 31(1), 99-107.
Jeong Hun Park, Young-Kyoon Suh and Jeeyoung Kim. 2026, "Early Classification of Parkinson’s Disease Using Mel-Spectrogram Voice Analysis via Transfer Learning", Journal of The Korea Society of Computer and Information, vol.31, no.1 pp.99-107.
Jeong Hun Park, Young-Kyoon Suh, Jeeyoung Kim "Early Classification of Parkinson’s Disease Using Mel-Spectrogram Voice Analysis via Transfer Learning" Journal of The Korea Society of Computer and Information 31.1 pp.99-107 (2026) : 99.
Jeong Hun Park, Young-Kyoon Suh, Jeeyoung Kim. Early Classification of Parkinson’s Disease Using Mel-Spectrogram Voice Analysis via Transfer Learning. 2026; 31(1), 99-107.
Jeong Hun Park, Young-Kyoon Suh and Jeeyoung Kim. "Early Classification of Parkinson’s Disease Using Mel-Spectrogram Voice Analysis via Transfer Learning" Journal of The Korea Society of Computer and Information 31, no.1 (2026) : 99-107.
Jeong Hun Park; Young-Kyoon Suh; Jeeyoung Kim. Early Classification of Parkinson’s Disease Using Mel-Spectrogram Voice Analysis via Transfer Learning. Journal of The Korea Society of Computer and Information, 31(1), 99-107.
Jeong Hun Park; Young-Kyoon Suh; Jeeyoung Kim. Early Classification of Parkinson’s Disease Using Mel-Spectrogram Voice Analysis via Transfer Learning. Journal of The Korea Society of Computer and Information. 2026; 31(1) 99-107.
Jeong Hun Park, Young-Kyoon Suh, Jeeyoung Kim. Early Classification of Parkinson’s Disease Using Mel-Spectrogram Voice Analysis via Transfer Learning. 2026; 31(1), 99-107.
Jeong Hun Park, Young-Kyoon Suh and Jeeyoung Kim. "Early Classification of Parkinson’s Disease Using Mel-Spectrogram Voice Analysis via Transfer Learning" Journal of The Korea Society of Computer and Information 31, no.1 (2026) : 99-107.
1