Emotions are a part of human communication. Mood, feelings and personality of any human being can be recognized through his emotions. People have always found music significant in their lives. Music is a language of an emotion. It frequently expresses emotional qualities and qualities of human personality such as happy, sadness, aggressiveness, tenderness etc. It has a central role in human society because it so strongly evokes feelings and affects social activities and interactions.

In this project, we will explore Song Lyrics as well as TED Talks Analysis by focusing on the simple yet non-trivial task of categorizing them by emotional polarity. Here we will be applying natural language processing for cleaning, structuring and preprocessing data. Classification itself will be done using machine learning algorithms.

Lyrics provide high-level information about a song. We can gather this meta-information such as the genre, sentiment, and theme of a song simply by reading its lyrics. However, automating this task is very difficult. Automatic music classification systems typically focus on classification by audio features or collaborative filtering. However, these methods have numerous drawbacks. Audio feature processing relies on having the actual recording of a song, which may be difficult to obtain under copyright laws. Hence, we would be automating songs and talks classification using their lyrics and captions respectively.