An Efficient SMOTE-Based Deep Learning Model for Voice Pathology Detection

Abstract

The Saarbruecken Voice Database (SVD) is a public database used by voice pathology detection systems. However, the distributions of the pathological and normal voice samples show a clear class imbalance. This study aims to develop a system for the classification of pathological and normal voices that uses efficient deep learning models based on various oversampling methods, such as the adaptive synthetic sampling (ADASYN), synthetic minority oversampling technique (SMOTE), and Borderline-SMOTE directly applied to feature parameters. The suggested combinations of oversampled linear predictive coefficients (LPCs), mel-frequency cepstral coefficients (MFCCs), and deep learning methods can efficiently classify pathological and normal voices. The balanced datasets from ADASYN, SMOTE, and Borderline-SMOTE are used to validate and evaluate the various deep learning models. The experiments are conducted using model evaluation metrics such as the recall, specificity, G, and F1 value. The experimental results suggest that the proposed voice pathology detection (VPD) system integrating the LPCs oversampled by the SMOTE and a convolutional neural network (CNN) can effectively yield the highest accuracy at 98.89% when classifying pathological and normal voices. Finally, the performances of oversampling algorithms such as the ADASYN, SMOTE, and Borderline-SMOTE are discussed. Furthermore, the performance of SMOTE is superior to conventional imbalanced data oversampling algorithms, and it can be used to diagnose pathological signals in real-world applications.

Dual Attention Convolutional AutoEncoder for Diagnosis of Alzheimer’s Disorder in Patients Using Neuroimaging and MRI Features

“…The sole metric employed to assess accuracy in this study. The primary aim of the study [15] was to provide a fast and efficient diagnostic approach for identifying people who are in a state of good health before the occurrence of moderate cognitive impairment. In addition, this work introduces a new method for identifying the stages of Alzheimer's disease by integrating tensor-based morphometric image analysis with machine-learning architectures.…”

“…Their vision transformer achieved 81.25% accuracy. Xception architecture was utilized for the classification of Alzheimer's disease [15].…”

Simulation and recognition technology for infrared target imaging

“…Typically, only a limited amount of real-world sample data is available, leading to the problem of small sample size. To address the issue of small sample size, data augmentation techniques 5 have been employed to enhance the diversity of data. However, methods such as data augmentation through rotation, color variation, and oversampling may distort infrared information.…”

Multitask and Transfer Learning Approach for Joint Classification and Severity Estimation of Dysphonia

“…Several studies have been conducted in the research area of voice disorder. One study achieved an accuracy of 98.9% using CNN with oversampling techniques to deal with class imbalance [30] . CNN with chromagram feature set in [31] , CNN combined with RNN in [32] .…”