Bangla Speech Baesd Emotion Detection

Abstract

When humans communicate with each other, emotion plays quite a major role. Itallows us to consider the emotions of others and give them meaningful responses. Since it is a subjective matter, identification of emotion is very difficult. Here, we will be trying to develop a system called ”Bangla Speech Based Emotion Detection”, which will analyze and classify the embedded emotions. Our system will convert audio data to text and perform emotion detection on the text. As we can categorize the human emotion state in 5 different categories: Excited, Happy, Neutral, Sad and Angry our system will predict the emotional state from the textual data and categorize them accordingly. For that reason, we are creating datasets based on audio files which contain Bangla sentences. We will convert those audio files into textual data. After the conversion, we will apply Bert algorithm to it for our emotion detection.