Machine Learning based Gender Prediction by Exploring Diverse Audio Features
Zar Khanam, Muhammd Dawood Semab, Hikmat Ullah Khan, Malik Muhammad Ali Shahid · 2023
Gender prediction from diverse types of data is an important research domain thanks to its vast applications. In this research study, our focus is to analyze audio data, carry out exploratory data analysis and then use diverse audio features for prediction of diverse machine learning and deep learning algorithms. For empirical analysis, conventional machine learning algorithm such as Support Vector Machines and Logistic Regression, ensemble learning based algorithm of Random Forest and Deep learning-based model of Convolutional neural network has been applied on standard datasets in the relevant literature. Furthermore, we explored the impact of different audio feature sets, including Mel-frequency cepstral coefficients (MFCCs) and spectrograms, on the performance of the classification models, aiming to identify the most informative features for accurate gender prediction. The results are evaluated using standard performance evaluation measures such as accuracy, precision, recall and f-measure. Deep learning-based algorithm outperforms other algorithms and achieves accuracy of 89%.