Emotion Variance in Speech and Text

Emotion recognition is an area of research that is gaining increasing attention and may provide some useful insights. A speech signal is used to determine the emotional state of an individual through Speech Emotion Recognition (SER). The motive of the project is to understand and interpret emotional cues in speech and text, which is an important skill for effective communication and social interaction[1]. A multiclass classification problem is formalized as our problem and two models are compared based on their performance. For speech, we extracted acoustic and spectral features from the audio signals. For text, Learning-based methods are being used to predict emotions. While keyword-based detection methods use keywords to detect emotions, learning-based methods apply various machine learning theories to a previously trained classifier in order to detect emotions such as the Multinomial Naive Bayes Classifier (MNB). The input text will be categorized according to its emotion. Overall, we compare the emotion predictions from both speech and text and visualize their coincidence using R