The input originates as audio, our final model operates entirely on text. Whisper transcribes speech into text, and our classifier performs NLP-based formality analysis. Therefore, the core of our project is Natural Language Processing, specifically transformer-based text classification.