A Study of Emotion Classification of Music Lyrics using LSTM Networks

Emotion Recognition is a vital component of human-computer interaction and plays a pivotal role in applications such as sentiment analysis, virtual assistants, and affective computing. Long Short-Term Memory (LSTM) models are a subset of Recurrent Neural Networks (RNNs). It has gained significant popularity for their effectiveness in sequence modeling tasks, including emotion recognition. The study presents a review on the application of Long Short-Term Memory (LSTM) networks for emotion classification using music lyrics. It offers a thorough review of relevant literature and outlines the methodology for implementing LSTM models for emotion recognition. Furthermore, the study emphasizes the significance of hyperparameter tuning in building effective machine-learning models, particularly LSTM-based models.

I. Introduction

Music Lyrics are artists' instruments to express their emotions, thoughts, and experiences. Songwriters can communicate emotions through carefully chosen words and phrases like love, happiness, sadness, anger, and nostalgia. Music streaming platforms can create playlists based on the emotion present in lyric text. For example, they can curate playlists for "calm," "sad," "relaxing," or "summer vibes," selecting songs that share similar emotions in their lyrics. It allows users to find and listen to music that aligns with their current emotional state or desired mood. The personal nature of lyrics enables listeners to connect with the conveyed emotions.