Enhancing Bangla speech emotion recognition

Abstract

Speech Emotion Recognition (SER) has grown to be a crucial part of humancomputer interaction and other advanced speech recognition systems. Although Bangla is one of the most widely spoken languages in the world, there has been a lack of extensive research on Bangla speech emotion recognition. Speech, a complex analogue signal that varies over time, presents unique challenges and opportunities for researchers implementing machine learning techniques in audio signal classification. In this research we will introduce a new benchmark Bangla Speech Emotion recognition dataset. We will also develop a benchmark method to effectively comprehend emotion recognition from Bangla speech. The method will include complex deeplearning techniques to classify emotion from the audio speech. The research topic proposes many benefits such as the potential for developing more emotionally aware machine interactions in diverse and low-resource linguistic contexts. The broader implications encompass the advancement of human-computer interaction technologies, rendering them more inclusive and proficient in comprehending a broader spectrum of emotional expressions, including mixed emotions.