This website implement the Bidirectional Long Short Term Memory (Bi-LSTM) algorithm for the classification of emotions in song lyrics, implement word embedding of Word2Vec, GloVe, and FastText, and identify which parameters influence the application of word embedding model and Bi-LSTM algorithm for classifying emotions in song lyrics.

• Developed a web application using Flask to classify emotions in song lyrics, providing an accessible platform for testing deep learning models in real-time.
• Implemented natural language preprocessing pipelines, improving text data quality and boosting model training efficiency by 60%.
• Integrated three word embedding models (Word2Vec, GloVe, and FastText) under consistent settings, enabling systematic comparison and identifying the most effective embedding for emotion detection.
• Applied a Bidirectional Long Short-Term Memory (BiLSTM) algorithm, achieving improved classification accuracy and delivering insights into parameter influences on emotion analysis, supporting further research in NLP for music and sentiment studies.