A Hybrid Ensemble Stacking Model for Gender Voice Recognition Approach

Gender recognition by voice is a vital research subject in speech processing and acoustics, as human voices have many remarkable characteristics. Voice recognition is beneficial in a variety of applications, including mobile health care systems, interactive systems, crime analysis, and recognition systems. Several algorithms for voice recognition have been developed, but there is still potential for development in terms of the system's accuracy and efficiency. Recent research has focused on combining ensemble learning with a variety of machine learning models in order to create more accurate classifiers. In this paper, a stacked ensemble for gender voice recognition model is presented, using four classifiers, namely, k-nearest neighbor (KNN), support vector machine (SVM), stochastic gradient descent (SGD), and logistic regression (LR) as base classifiers and linear discriminant analysis (LDA) as meta classifier. The dataset used includes 3168 instances and 21 features, where 20 features are the predictors, and one feature is the target. Several prediction evaluation metrics, including precision, accuracy, recall, F1 score, and area under the receiver operating characteristic curve (AUC), were computed to verify the execution of the proposed model. The results obtained illustrated that the stacked model achieved better results compared to other conventional machine learning models. The stacked model achieved high accuracy with 99.64%.