Analysis of sentiment is a pivotal component of natural language processing and has recently observed noteworthy evolutions. Still, the impact of gender on expressivity of the emotion remains an undiscovered area. The proposed work utilizes a broad range of over 12000 audio data samples from four different benchmarked datasets RAVDESS, CREMA-D, TESS, and SAVEE. Convolutional Neural Network (CNN) is used to identify and detect patterns and biases according to gender. The study found mixed-gender emotion accuracy at 84.26%, female emotion accuracy at 89.40%, and male emotion accuracy at 82.70%. The proposed work aims to demonstrate that the female voice is more expressive of emotion than the male voice by examining the difference in sentiment expression between genders. This research will enhance the insights of sentiment analysis and can be useful to contrivance industries ranging from customer service to human-system interaction.
Bapat et al. (Mon,) studied this question.