Time-Frequency Analysis of Speech Signal Using Wavelet Synchrosqueezing Transform for Automatic Detection of Parkinson's Disease

This letter proposes a novel method for detecting Parkinson's disease (PD) based on a time-frequency representation matrix (TFRM) of the speech signal generated by the wavelet synchrosqueezing transform (WSST). The energy and entropy of each frequency component of the TFRM are calculated and used as features for detecting PD using speech signals. Then, the genetic algorithm along with support vector machine (SVM) and gradient boosting models are utilized for classification. The results indicate that the proposed approach effectively detects PD using speech signals. We have obtained the maximum accuracy of 95% using the word /apto/. The proposed work shows better results in comparison to the majority of the existing state-of-the-art techniques.