Unveiling Gender from Speech: An Investigation into Acoustic Features for Accurate Gender Detection
Sandhya Samant, Aditi Aditi, Shobhit Prajapati, Renu Chaudhary, Aaditya Rathi, Swati Arya · 2023
This paper focuses on the development of a classification model for gender recognition based on voice speech using various acoustic parameters. The aim is to predict whether a voice sample belongs to a male or female speaker. The initial step in the paper involves dividing the dataset into two parts: training data and test data, through a test-train data split. Following this, the study explores the performance of various machine learning algorithms for gender classification based on voice features. The algorithms under consideration include Support Vector Machine (SVM), Decision Tree Classifier, k-Nearest Neighbors (KNN), Random Forest, and Logistic Regression. Each of these algorithms is individually trained on the training data and subsequently evaluated on the test data to gauge their effectiveness in gender classification. The paper then analyzes and compares the results obtained from each algorithm, assessing their accuracy, precision, recall, and F1-score to identify the model that performs best in predicting gender based on voice features. The performance of every model is assessed using a confusion matrix and classification report, offering valuable information about accuracy, precision, recall, and F1-score for each gender class. The model with the highest accuracy is determined as the best performer among them. By employing these classification algorithms and analysing the acoustic parameters of voice speech, this paper aims to provide a reliable method for gender recognition. The results obtained from this study can have potential applications in various fields, including speech processing, voice-based user interfaces, and automatic gender identification systems.