Detecting and Analyzing the Emotional Levels of a Person Through CBT Using MFCC and Lexicon-Based Approach

Depression is a serious illness that usually goes ignored by friends, family, and society. Despite not being recognized as such mental illness, depression is a severe mental condition that requires careful management. The AI industry has been obliged to take matters under its control due to cost and availability issues. As part of this study, our chatbot makes use of the Librosa library to evaluate audio and music while also listening for the speaker’s tone and voice. The lexicon-based approach is one approach for semantic and sentiment analysis from text. This technique establishes the sentiment orientations of the text. Both human and automated methods can be used to create the dictionary of lexicons. One way for extracting emotions from text is lexicon-based methods. Six distinct emotions may be recognized, and it contains certain predetermined phrases. A little amount of information from the voice signal is taken out using the feature extraction technique, which may then be utilized to identify each speaker individually. Using the Mel Frequency Cepstral Coefficient (MFCC) approach, an individual’s mood may be inferred from their voice. The effectiveness of the MFCC system was found to be roughly 80% when the happy, sad, and angry emotions were tested. By asking open-ended inquiries like “How are you?” and “How was your day?” the chatbot carries on the conversation and gathers additional data to better understand the user. We will employ datasets for voice recognition like Ravdess and Tess in order to train the AI model using a variety of emotions and words.