Sentiment analysis on music via deep learning

Abstract

This study delves into the domain of sentiment analysis, a facet of natural language processing (NLP) that discerns the positivity, negativity, or neutrality of textual data, to extend its application towards understanding emotional expressions in music. With the advent of deep learning, a paradigm shift in artificial intelligence (AI) has enabled the emulation of human brain functionalities through sophisticated neural network architectures. This research leverages deep learning methodologies, including Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN), Long Short-Term Memory (LSTM) networks, and Bidirectional Encoder Representations from Transformers (BERT), to analyze mood trends in Billboard Top 100 songs spanning the past five decades. Amidst the backdrop of the COVID-19 pandemic, which has significantly increased music consumption and highlighted music's role as a universal language, this study aims to unravel the evolution of emotional expressions in music. By examining the interplay between societal shifts, cultural influences, and emotional expressions within various music genres, it seeks to uncover patterns and emotional shifts that can offer invaluable insights. The objective is to bridge the gap between digital signals and human emotions, benefiting music consumers and providers alike by tailoring content more effectively to audience preferences. This research is motivated by the recognition that, despite the abundance of studies in sentiment analysis, there is a conspicuous lack of focus on the music industry and an imperative need for updated and implicative analyses, given the outdated datasets currently available for such studies.