Ensemble Learning Model for Gender Recognition Using the Human Voice

Gender recognition through the human voice is a rapidly growing field of research. It is a vital research subject as the human voice now has an ever-growing list of applications, especially for social good. Gender recognition using voice is beneficial in multiple applications such as online healthcare systems, interactive services, crime analysis, security systems, etc. Algorithms have already been deployed, but accuracy and efficiency can still be improved. This work aims to classify human gender based on human voice data between males and females. It uses modified ensemble techniques based on classifiers such as k-NN (k-nearest neighbors), Random Forest (RF), and SVM (support vector machine). In this study, a benchmark dataset is used which includes 3168 instances and 21 attributes, where 20 attributes are the predictors, and one attribute is the target – ‘male’ or ‘female’ as instances. To evaluate the proposed model’s results, precision, accuracy, recall, and F1-score were calculated. When compared to traditional machine learning models employed independently, the ensemble model imposed better results, with an accuracy of 99.05%.