Acoustic Methodologies for Classifying Gender and Emotions using Machine Learning Algorithms

The machine learning classifiers are found to be the most relevant classifiers regarding emotions and gender recognition using speech. In this research, emotions and gender are recognized with the help of two machine learning classifiers: Support Vector Machine and Naive Bayes which incorporates four speech features: shimmer, jitter, energy, and pitch for classification. The classifiers take these input sequences and correctly classify the gender as well as the emotion in the speech signal. Results shows, regarding classification accuracy, Support Vector Machine outperforms the Naive Bayes classifier by almost 10% in gender recognition and 35% in emotion recognition. The emotion and gender recognition techniques are applying in many telecommunication and security-based applications such as sentiment analysis, speech analysis, and other security purposes.